Crossroadsxrds2014winter DL

XRDS
Crossroads The ACM Magazine for Students
W IN T ER 2014 V OL .2 1 NO.2
XRDS.ACM.ORG
Health
Informatics
The Anatomy of a
Human Disease Network
Did I Take My Meds Today?
Challenges in Personal
Health Tracking
ACMs Career
& Job Center
Are you looking for

your next IT job?
Do you need Career Advice?
The ACM Career & Job Center offers ACM members a host of
career-enhancing benefits:
A highly targeted focus on job

opportunities in the computing industry
Job Alert system that notifies you of

new opportunities matching your criteria
Access to hundreds of industry job postings
Resume posting keeping you connected

to the employment market while letting you
maintain full control over your confidential
information
Career coaching and guidance available

from trained experts dedicated to your
success
Free access to a content library of the best

career articles compiled from hundreds of
sources, and much more!
Visit ACMs
Career & Job Center at:

http://jobs.acm.org
The ACM Career & Job Center is the perfect place to
begin searching for your next employment opportunity!
Visit today at http://jobs.acm.org
Inviting Young
Scientists
Meet Great Minds in Computer
Science and Mathematics
As one of the founding organizations of the Heidelberg Laureate Forum http://
www.heidelberg-laureate-forum.org/, ACM invites young computer science and
mathematics researchers to meet some of the preeminent scientists in their field.
These may be the very pioneering researchers who sparked your passion for
research in computer science and/or mathematics.
These laureates include recipients of the ACM A.M. Turing Award, the Abel Prize,
and the Fields Medal.
The Heidelberg Laureate Forum is August 2328, 2015 in Heidelberg, Germany.
This week-long event features presentations, workshops, panel discussions, and
social events focusing on scientific inspiration and exchange among laureates and
young scientists.
Who can participate?
New and recent Ph.Ds, doctoral candidates, other graduate students

pursuing research, and undergraduate students with solid research
experience and a commitment to computing research
How to apply:
Online: https://application.heidelberg-laureate-forum.org/
Materials to complete applications are listed on the site.
What is the schedule?
Application deadlineFebruary 28, 2015.

We reserve the right to close the application website
early depending on the volume
Successful applicants will be notified by April 15, 2015.
More information available on Heidelberg social media
PHOTOS: HLFF/B. Kreutzer (2)
Crossroads
The ACM Magazine for Students
W IN T E R 2 014 V OL . 2 1 NO . 2
12
begin
5 LETTER FROM THE EDITORS
7 INBOX
8 INIT
HEALTH 2.0:
The Digital Health Revolution
By Diana Lynn MacLean
9 ADVICE
The Essentials of a
Computer Scientists Toolkit
By Numair Khan
10 UPDATES
Staying In Touch
By Claudia Schulz
11 MILESTONES
Vital Technology
By Jay Patel
12 CAREERS
14 BLOGS
My $300 Home Cloud Server:

A Story of Blood, Sweat, and eBay
By Wolfgang Richter
X R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
Image by Valentina Photo / Shutterstock.com
An Antidote to Impostor Syndrome

By Dean Jackson and Taliver Heath
Health Informatics
23
44
features
18 FEATURE
Gathering People to Gather Data

23 FEATURE
Opportunities of Social Media

in Health and Well-Being
By Munmun De Choudhury
62
end
38 FEATURE
Did I Take My Meds Today?

By Matthew L. Lee
56 LABZ
The Regenstrief Global Health

Informatics Group
By Suranga Nath Kasthurirathne
44 FEATURE
Seeing Is Believing
By Megan Monroe
57 BACK
Radiography
By Finn Kuusisto
48 FEATURE
Left Image by Everything Possible / Shutterstock.com; Right Image by Sean Pavone / Shutterstock.com
28 FEATURE
Here Comes the #Engagement:

A serious health initiative made
trendy
By Fay Cobb Payton and
KaMar Galloway
Wearable Technologies: One step

closer to gait rehabilitation in
Parkinsons patients
By Sinziana Mazilu
and Gerhard Trster
58 HELLO WORLD
The Anatomy of
a Human Disease Network
By Marinka Zitnik
61 POINTERS
54 PROFILE
32 FEATURE
Challenges in Personal Health

Tracking: The data isnt enough
By Matthew Kay
Trevor van Mierlo: The Story

of Building a Startup
in Health Informatics
By Adrian Scoica
61 ACRONYMS
62 EVENTS
64 BEMUSEMENT
We would like to acknowledge former XRDS feature
editor Virginie Lerays, who was not properly credited
in the last issue. She worked alongside Daniel Bauer
and Adrian Scoic on the Natural Language issue.
M.S. programs at
E DI T ORI A L B O A RD
Editors-in-Chief
Inbal Talgam-Cohen
Stanford University, USA
Sean Follmer
MIT, USA
Your future career, secured.

Master the technologies, economics and policies
of secure communication networks, systems and
services
Examples of research areas:
big data analytics
browser security
cloud computing
Internet of Things
machine learning
network security
operating systems
sensor networks
storage and file systems
wireless systems
network architecture
and more
Full federal scholarships available for U.S. citizens
www.ini.cmu.edu
Departments Chief
Vaggelis Giannikas
University of Cambridge,
UK
Issue Editor
Diana MacLean
Stanford University, USA
Laurie Faith Cranor,

Carnegie Mellon
Alan Dix,
Lancaster University
David Harel,
Weizmann Institute
of Science
Bill Stevenson ,
Apple, Inc.
Erin Claire Carson

University of California
Berkeley, USA
Noam Nisan, Hebrew

University Jerusalem
Andrew Tuson,
City University London
Richard Gomer
University of
Southampton, UK
Jeffrey D. Ullman,
InfoLab, Stanford
University
Suranga Kasthurirathne
Indiana UniversityPurdue University, USA
Moshe Y. Vardi,
Rice University
Numair Khan
New York University, USA
E DI T ORI A L S TA F F
Director, Group
Publishing
Scott E. Delman
Talia Kohen
Bar Ilan University,Israel
Hanieh Moshki
HEC Montreal , Canada
Somdip Dey
University of
Manchester, UK
Rohit Goyal
West Chester East High
School, USA
Bryan Knowles
Western Kentucky
University, USA
Finn Kuusisto
University of
Wisconsin-Madison, USA
Jay Patel
University of California
Berkeley, USA
Ashok Rao
University of
Pennsylvania, USA
Claudia Schulz
Imperial College London, UK
Adrian Scoica
University of Cambridge,
UK
Apoorvaa Singh
IPEC Ghaziabad, India
Marinka Zitnik
University of Ljubljana,
Slovenia
Digital Content Editor
Pedro Lopes
Hasso Plattner Institut,
Germany
Web Editors
Shelby Solomon Darnell
Clemson University, USA
Bernard Chazelle,
Princeton University
Feature Editors
Arka Bhattacharya
Columbia University, USA
Students and faculty can take advantage

of ACMs Distinguished Speakers Program
to invite renowned thought leaders in
academia, industry and government to
deliver compelling and insightful talks
on the most important topics in computing
and IT today. ACM covers the cost of
transportation for the speaker to travel
to your event.
Mark Allman,
International Computer
Science Institute
Panagiotis Takis Metaxas ,

Wellesley College
Department Editors
http://dsp.acm.org
A D V IS OR Y B O A RD
Issue Feature Editor

Hanieh Moshki
HEC Montreal , Canada
Billy Rathje
University of Puget
Sound, USA
Distinguished Speakers Program
Poornima Sundaram
University of Southern
California, USA
XRDS Managing Editor &

Senior Editor at ACM HQ
Denise Doig
Production Manager
Lynn DAddessio
Art Direction
Andrij Borys Associates,
Andrij Borys,
Mia Balaquiot
Director of Media Sales
Jennifer Ruzicka
jen.ruzicka@acm.org
Copyright Permissions
Deborah Cotton
permissions@acm.org
Public Relations
Coordinator
Virginia Gold
ACM
Association for
Computing Machinery
2 Penn Plaza,
Suite 701
New York, NY
10121-0701 USA
+1 212-869-7440
C ON TA C T
General feedback:
xrds@acm.org
For submission
guidelines, please see
http://xrds.acm.org/
authorguidelines.cfm
SUB S C RIBE
Subscriptions ($19
per year includes XRDS
electronic subscription)
are available
by becoming an
ACM Student Member
www.acm.org/
membership/student
Non-member
subscriptions:
$80 per year
http://store.acm.org/
acmstore
ACM Member Services
To renew your ACM
membership or XRDS
subscription, please send
a letter with your name,
address, member number
and payment to:
ACM General Post Office
P.O. Box 30777
New York, NY
10087-0777 USA
Postal Information
XRDS (ISSN# 1528-4981)
is published quarterly in
spring, winter, summer
and fall by Association for
Computing Machinery,
2 Penn Plaza, Suite 701,
New York, NY 10121.
Application to mail at
Periodical Postage rates
is paid at New York, NY
and additional mailing
offices.
POSTMASTER: Send
addresses change to:
XRDS: Crossroads ,
Association for
Computing Machinery,
2 Penn Plaza, Suite 701,
New York, NY 10121.
Offering# XRDS0171
ISSN# 1528-4972 (print)
ISSN# 1528-4980
(electronic)
Copyright 2014 by the
Association for Computing
Machinery, Inc. Permission
to make digital or hard
copies of part of this work
for personal or classroom
use is granted without fee
provided that copies are
not made or distributed
for profit or commercial
advantage and that copies
bear this notice and the
full citation on the first
page or initial screen of
the document. Copyrights
for components of this
work owned by others than
ACM must be honored.
Abstracting with credit
is permitted. To copy
otherwise, republish, post
on servers, or redistribute
requires prior specific
permission and a fee.
Permissions requests:
permissions@acm.org.
PUBLIC AT IONS BOA RD

Co-Chairs
Jack Davidson and
Joseph A. Konstan
Board Members
Ronald F. Boisvert,
Marie-Paul Cani,
Nikil Dutt, Roch Guerrin,
Carol Hutchins,
Patrick Madden,
Catherine C. McGeoch,
M. Tamer Ozsu,
Mary Lou Soffa
LETTER FROM THE EDITORS
A Shrimps Tale:
Why we need
to fund research
F
or me, it all started with a YouTube video1 of a shrimp on an underwater treadmill
accompanied by the Benny Hill theme song. Why was this shrimp on an endless
journey, running seemingly forever, and why was it so gosh darn funny? The humor
question is easyyou dont see shrimp hitting the gym every day. But why did this video
exist in the first place? Of course there are thousands, if not millions, of humorous videos
involving treadmills on YouTube, but only this one was being investigated by the U.S. House of
Representatives Committee on Science, Space and Technology. It turns out this video was part
UPCOMING ISSUES
Summer 2015
[June issue]
Computational Biology
Article deadline: February 27, 2015
Fall 2015
[September issue]
Virtual Reality
Article deadline: July 17, 2015
of research that was funded by the

United States National Science
Foundation (NSF).
You might start to wonder why putting a shrimp on a treadmill is important research, but let me assure
you, it is. Professor David Scholnick at
the Pacific University in Oregon conducted this research to understand
how water quality can affect shrimps
movements and the distances they
can travel in a short time. And Pro1 https://www.youtube.com/watch?v=Qj-yAHTfVeE
fessor Robert Full, a biologist at UC

Berkeley who works closely with roboticists, uses similar videos of various animals (from crabs to cats) on
treadmills to understand the biomechanics of their locomotion. It turns
out these animals are extremely efficient. By understanding how nature
can achieve such highly efficient locomotion, we may be able to build more
advanced and smaller robots. New,
efficient mini robots could be deployed to search and rescue missions
because they are smaller and could
run for longer on a battery.
5
We need research
that not only
points to new,
more efficient
algorithms,
but also to
more efficient ways
to use computing
to stop the spread
of disease or
to help people
lose weight.
This formative basic research may

seem unimportant or even comical,
yet it has the potential to provide discoveries that could change the way
robotics works or have other unforeseen applications.
Why should we do this basic research and more importantly why
should anyone, especially the government, pay for it? And why should
computer science research take place
in academia if there are thousands
of companies in Silicon Valley and
around the world creating new, innovative technologies and products?
These questions are becoming increasingly important for us as a research community to answer as interest in federal funding for the sciences
decreases.
Other fields have long dealt with
having to provide financial justification for their research. Physicist David Kaplan, theoretical particle physics expert from John Hopkins and
producer of the film Particle Fever
6
about the Large Hadron Collider at

CERN, often gets asked about evaluating the results of such a large project
in financial terms. His reply: I have
no idea. We have no idea. When radio
waves were discovered, they werent
called radio waves, because there
were no radios. They were discovered
as some sort of radiation.2
To me this is at the heart of why basic research is so important, because
undiscovered knowledge can impact
the world in amazing, unintended,
and unforeseeable ways. Often advanced research topics may sound obscure or arcane, or even unnecessary
(much like a shrimp on a treadmill)
to outsiders, but experts in the field
may think these are the very projects
that should get funded. Right now in
the U.S. a witch-hunt, reminiscent of
the Golden Fleece award of the 1970s
and 80s, is on to find government
funded research that should not have
been funded. Currently the House
Committee on Space and Science has
been investigating a number of NSF
research grants across a wide variety
of topics, including computing projects that investigate crowd sourcing,
ubiquitous sensing on mobile devices, and creativity support tools and
gaming.3 And the time-honored tradition of anonymous expert peer review of research projects at the NSF is
coming under close scrutiny as well.
Of course, there are many issues with
peer review, and we should no doubt
have more oversight into how the NSF
allocates funds, but my fear is that
these investigations are very politically motivated.
In many ways this is part of a larger goal of the House Committee on
Space and Science to defund social
science research.4 While we as computer science students may think this
has little to do with us, it actually hits
much closer to home. Computing
research is increasingly focused on
2 http://sploid.gizmodo.com/heres-theperfect-answer-to-why-we-must-invest-inpure-1665820877/+ericlimer
3 http://news.sciencemag.org/policy/2014/10/
battle-between-nsf-and-house-science-committee-escalates-how-did-it-get-bad
4 https://www.insidehighered.com/
news/2014/06/02/house-passes-nsf-fundingbill-takes-slap-social-sciences
social networks and understanding

how people interact with and through
computers. We need research that not
only points to new, more efficient algorithms, but also to more efficient
ways to use computing to stop the
spread of disease or to help people
lose weight. This issue covers health
and technology, and what we see is
that so much of this research really
focuses not only on computing, but
also on understanding people and the
intersection between CS and social
science, or CS and health or biology.
This kind of research is critical for CS
and for CS students to have an impact
in the world today. However, it seems
we as a community need to do a better job of conveying the importance
and value of our work. In the age of
YouTube, TED, and crowd-sourced
funding on Kickstarter, we need to
be thinking beyond our publication
record.
One could make an argument that
CS research could be done by the
large tech industry. Certainly there
is great innovation coming from industry in the form of new technologies, new applications, and also new
research topics (Microsoft Research
is one of the largest publishers of CS
research). But, often the constraints
that help industry come up with new
solutions can be burdensome for
pursuing research that has no clear
financial benefit today. And the fast
pace of industry means labs come
and go,5 interest in research topics
change quickly, and internal funding
is often short term, making it hard to
have long-term research projects. We
still need publicly funded research in
computing.
Social science and applied research matter. And they matter in a
CS context. We student members of
the computing community need to
stand up and make it clear that we
see a future in which computing is an
integral part of any research agenda,
ensuring that computing has a broad
and diverse definition in the years to
come.
5 http://www.zdnet.com/article/microsoft-toclose-microsoft-research-lab-in-silicon-valley/
Sean
Follmer and
Inbal Talgam-Cohen
begin
INBOX
ON THE NLP ISSUE

For #NLP people: did anyone find the easter egg we
hid on the cover of the current issue of @XRDS_ACM
(which is about Natural
Language)?
Daniel Bauer, Natural
Language Processing Ph.D.
candidate, Columbia
University, Twitter (@
dnlbauer)
My article From wax
tablets to touchscreens: an
introduction to text entry
research is out in latest @
XRDS_ACM: http://pokristensson.com/pubs/KristenssonXRDS2014.pdf
Per Ola Kristensson,
University Lecturer
Department of
Engineering, University
of Cambridge, Twitter (@
pokristensson)
Image by KreativKolors
NEW ON THE BLOG

Why and How #Parallel#
Programming? A great
post by Fahad at @XRDS_
ACM http://xrds.acm.org/
blog/2014/09/parallelprogramming-throughdependence-analysispart-i/ #computing
Pedro Lopes, Hasso
Plattner Institut, HCI
researcher, Twitter
(@plopesresearch)
A QUESTION
ABOUT EFFICIENCY
Have you connected
something like a kill-awatt to see what the power
draw is? I think youll find
that its pulling an absurd

amount of power. As a
general rule of thumb, each
watt is $1 to run 24/7. That
means if its pulling 300
watts as I suspect, youre
paying an extra $200/year
more than more energy
efficient hardware. It
may still be worth it, but
I wanted to point out the
hidden cost. Sometimes
its worth replacing old
hardware just for the power
savings from
a financial aspect.
Jason Young,
Azure Developer
Bloggers Reply:
Totally true. But for our use
case, research workloads, we
only turn it on sporadically
(2-4 times per week).
Wolfgang Richter,
Graduate Student, School
of Computer Science,
Carnegie Mellon University
OTHER TWEETS
@XRDS_ACM
@marinkazitnik
My friends became
wizards on MUDs (http://
en.wikipedia.org/wiki/
MUD ) and had to do magic
by writing code.
They learned fast!
Andrej Bauer,
Mathematician,
computer scientists,
Twitter (@andrejbauer)
How to contact XRDS: Send a letter to the

editors or other feedback by email (xrds@
acm.org), on Facebook by posting on our
group page (http://tinyurl.com/XRDSFacebook), via Twitter by using #xrds
in any message, or by post to ACM Attn:
XRDS, 2 Penn Plaza, Suite 701, New York,
New York 10121, U.S.
In the U.S., although medical records contain patient

information, medical providersincluding hospitalsown
all patient records, which can be shared with insurance
companies, pharmacies, researchers, and employers.
INIT
ts the crack of dawn,

and Jane is on her
morning run with her
smartphone. She uses
a running app to pace herself through the full six
miles without stopping.
She notices her split times
are a bit slower than usual,
and wonders if it has something to do with her restless night; she had a tough
time falling asleep. When
she gets home, Jane examines the charts her wireless
scale generates. Over the
past two weeks, her body fat
percentage has dropped,
and her weight has gone
up: Shes built some muscle. Looks like incorporating those strength sessions
into her morning workouts
is paying off!
While its hard today to
imagine a world in which
technology and health are
not enmeshed, the phenomenon is somewhat recent.
The tech + health (or Health
2.0) movement is rooted in
the mid 90s, with the advent of the commercial web.
The Internet made medical
information available to
anyone behind a browser:
No longer did people have
to overcome financial, physical, or geographic barriers to learn about diabetes
treatments, for example, or
to research Lyme Disease
symptoms. Since then, other technological advances
(e.g., the increasing affordability and sophistication

of sensor technology) have
only served to more deeply
embed Health 2.0 into our
everyday environments.
Beyond ubiquity, however, Health 2.0 has dramatically reformed the
role patients play in their
own healthcare. Just like
Jane, many of us are active
participants in managing
our own health: We turn to
technology to help us lose
weight, improve long-term
memory capacity, figure out
what triggers our asthma
attacks, and to collaborate
and share experiences with
people who have the same
health concerns as us. And
as a result of our participation in Health 2.0, healthrelated datawhether in
The idea that

we can leverage
technology
on novel data
to improve
both our own
well-being as
well as that of
those around us
is compelling.
the form of plain text, survey responses, or even raw

biometric datais both
abundant and easier to collect, store, and share than
ever before.
In the past few decades,
Health 2.0 has grown into
one of the most vibrant
and active topics in computer science, and it is easy
to intuit why. In a world in
which almost anything can
be measured, we have all
become, to some degree,
direct beneficiaries of the
Health 2.0 movement. The
idea that we can leverage
technology on novel data
to improve both our own
well-being as well as that of
those around us is compelling. What if, for example,
we could use social media
data to preemptively identify disease outbreaks? What
if visualizing patients medical histories intuitively
helped doctors make better

treatment decisions? What
if, armed with sensors and
analytics, we could crack
to code to getting a good
nights sleep, every night?
In this issue, we bring
you a glimpse of cuttingedge research that begins
to answer some of these
questions, and more. From
the inside scoop on online
health communities, to
the question of social medias role in diagnosing
and tracking conditions, to
technological innovations
for hacking our well-being, to visual representations of medical histories,
to designing and building
a gait training system for
people living with Parkinsons Disease, we have
packed in a collection of articles that represents the vibrance and excitingness of
Health 2.0 itself. Of course,
such excitement is accompanied by thoughtful caution. Issues of privacy, data
quality, and health policy
remain pertinent and in
tension with newer technological advances. We hope
to leave you with a piqued
interest, a balanced perspective, and most importantly, enthusiasm about
the current and future outlook of Health 2.0.
Diana
Lynn MacLean,
Issue Editor
Photo by Katherine Breeden
HEALTH 2.0:
The Digital Health Revolution
A data silo refers to records

owned and stored by one provider,
but not shared with others.
ADVICE
The Essentials of a
Computer Scientists Toolkit
Image by Slavoljub Pantelic
ho wants to set breakpoints when they can

pepper their code with
far more insightful statements like printf(asdafs)? While
learning how to use a debugger is a skill
most programmers love to hate, anybody who has spent long hours trying
to make sense of a crash reportoften
couched in equally insightful terms by
the operating systemknows better.
They say a wise person learns from the
experience of others and as I hold a very
high opinion of the mental capacity of
my readers, I have compiled the following list of skills that I was either thankful for possessing as a computer science
student, or, lacking which, was often
left pulling my hair in anguish.
Programming in C. Disregarding
the popular maxim that real men
(and women, of course) program in
C on the grounds that it, perhaps,
expresses a somewhat subjective and
unscientific opinion, the reasons for
learning this beautiful language are,
nevertheless, numerous. Not only is
it the progenitor of most modern programming languages, it provides the
perfect level of abstraction for understanding the architecture of the underlying machine. In my experience,
C programmers generally possess a
better ability to think low level and
have a more solid understanding of
core programming principles. Moreover, contrary to popular belief, C is
not a dead language. Nor is it difficult
to learn, being amongst the simplest
and most concise programming languages around.
Thinking parallel. For the past few
years, multiple cores rather than high
clock-speeds have driven microproX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
cessor performance and the trend is

expected to continue in the future.
This means sequential programs can
no longer hope for regular speedups
and all future programs must be,
to some degree, concurrent. Not so
much a practical skill as a state of
mind, the ability to think parallel is,
nonetheless, indispensable to programmers who wish to exploit the full
potential of multicore processors.
Knowledge of basic probability. As a
computer scientist, expect to encounter uncertainty often and be prepared
to analyze and quantify it. Familiarity with statistical inference methods
is especially important if you plan on
conducting research.
Unix. Computer clusters at academic institutions and large corporations commonly run Unix (or one
of its variants) and to take advantage of these resources you should
feel at home logging in, performing basic file operations, using the
compiler, and editing text files from
a Unix shell. Add the ability to write
simple scripts to the mix and before

you know it, youll have a Live Free
or Die license plate.
LaTeX. A typesetting language, LaTeX makes it extremely convenient to
produce high-quality scientific documents. Not only does it help you as an
editor, but also being the de facto standard for research publications, it gives
each document a well-established academic look.
The Make utility. This last choice
may seem a bit controversial. But,
believe me, if I had a nickel for every time my eager efforts to compile
downloaded source code have been
foiled by the appearance of a cursed
makefile. For the unfamiliar, make
is a utility for managing and building source files, and most online codebases include a makefile specifying the order of compilation and any
build parameters. If you intend to
interact with a large codebase and
maintain a good head of hair at the
same time, learn to use make.
Numair
Khan
9
begin
UPDATES
Staying In Touch
How print and digital media can
further student chapter outreach
This years
committee was even
more enthusiastic
about issuing
the newsletter and
spent a lot of effort
to perfect it .
10
The first two pages of the 2013/14 newsletter, introducing the head of the editorial department and the executive committee (left) and a preview of topics covered
in the newsletter (right).
which are advertised, carried out,

and documented.
The ACM SZABIST Student Chapter is very aware of the importance
of keeping in touch with the student
community, providing updates about
upcoming chapter activities, and
sharing advances in the field of computer science. This is why the Chapter
has a Social Media Department. Facebook is their most important communication platform. There are 483
Facebook group members; its page
has more than 1,000 likes. In addition, updates are also announced on
Twitter and Instagram, following a
current trend toward these new social media platforms. Its not very
surprising that email is rarely used to
spread information. People here do
use them [emails] but mostly for official work; so social media works much
better for us than emails, explained
chapter president Michael Fernandes.
During the 2011-2012 academic
year, the chapters Executive Body
introduced the idea of creating an
annual publication to share updates
about the chapter as well as news in
the world of computing. An Editorial
Department was thus formed, which
published the first issue at the end of
Photo Credit TK
n this edition of XRDS, I would like

to introduce you to a student chapter that is not only using digital
platforms, such as social media, to
reach students, but also flexing their
creative skills to raise chapter awareness using traditional forms of print
mediaa newsletter.
Established in 2007 at the Shaheed Zulfiqar Ali Bhutto Institute of
Science and Technology (SZABIST) in
Karachi, Pakistan, the ACM SZABIST
Student Chapter knows what a good
structural organization is: In addition to its Executive Bodyconsisting of the president, vice president,
secretary, and treasurerthe Chapter has a Governing Body made of the
12 department heads. Each department has between two and 10 members and is responsible for a specific
task, such as event management,
photography, marketing, IT, and web
development. All contribute on a voluntary basis, making it possible to
organize about nine events per year,
2010
The Cyborg Foundation is founded by Neil Harbisson, who is known as the

worlds first cyborg. Its mission is to physically apply technology to the human
body to not only enhance, but create, new senses and perceptions.
MILESTONES
Vital
Technology
the academic year. This 12 page magazine not only introduced the members of the previous and new committee and provided information about
past and upcoming chapter events,
but also informed readers about current trends in computer science. Unfortunately, the following year the
committee was faced with some internal problems and was unable to issue
a newsletter at the end of the year. As
a consequence, this years committee was even more enthusiastic about
publishing the newsletter and spent a
lot of effort to perfect it.
Inspired by their leader Sana Haider, members of the Editorial Department diligently got to work. Their
articles highlighted the Chapters
3-D printing workshop and student
conference on programming mobile
phone applications, as well as driverless cars and Windows 8. Supported
by the Creative Department, which
did an excellent job in structuring the
newsletter and designing the layout,
a hard copy of this second newsletter was issued to academics and an
online version was published for students. The feedback from both groups
was very positive, encouraging the
chapter to begin work on publishing
another newsletter next year.
Feeling inspired? Have a look at
the online version of the most recent
ACM SZABIST chapter newsletter:
http://issuu.com/alishbachapsi/docs/
acm_newsletter_2013-2014, or visit
the Chapters webpage for more information: http://szabist.acm.org/about/.
Claudia Schulz
Vital signs are measures of the bodys basic functions and

can give an assessment of overall physical health. The four
major vital signs: are body temperature, blood pressure,
heart rate, and respiratory rate. The technology utilized
to measure these signs is constantly evolving to be more
comfortable and precise. Here are just a few milestones in
the history of vital technology:
1612
Santorio Santorio designs the first

clinical thermometer. It consists of a
thermoscopea container filled with buoyant bulbsand
a numerical scale. It is designed to be placed in the patients
mouth.
1881
The first sphygmomanometer, better known

as the blood pressure meter, is invented in
Austria and consists of a rubber bulb filled with water and
a mercury column. The device measure blood pressure in
millimeters of mercury.
1939
Researchers in Germany report using red and

infrared light in an ear oxygen meter, which
works by measuring changes in tissue thickness and light
intensity. This device is an early predecessor to the pulse
oximeter, which is a common non-invasive device used today
to measure blood oxygen saturation.
2014
Smartwatches are gaining increasing

popularity and feasibility as a tool for
measuring vital signs. For example, one of the most
publicized devices, the Apple Watch, is equipped with a builtin heart rate sensor using infrared and photodiodes.
Jay
Patel
11
begin
According to trends in Twitter data,

people tweet about health most often
between the hours of 6 a.m. and 9 a.m.
CAREERS
An Antidote to Impostor Syndrome
STRATEGY #1: TRACK

YOUR ACCOMPLISHMENTS
Track your accomplishments in a format where you can review them over
time. The format is up to you. The logging doesnt have to be long form; something short and concise is likely better.
When in doubt, focus on the positive
here, and make sure to document your
successes, even the small ones. When
you record a success, also make sure to
pause and reflect. Dwelling on successes as they happen can help to build and
strengthen long-term motivation.
Logging also helps to build a
12
healthy system of validation. Feedback

from others is never fully informed, as
others will not have full knowledge on
what you have done and why. You have
this missing context, which can be especially useful later for evaluating negative feedback. Tracking can help you
hold an honest light to criticism.
STRATEGY #2: COMPARE WITH CARE

Theres a related cognitive bias to impostor syndrome, called the Dunning
Kruger effect. Quoting from the initial
study, the miscalibration of the incompetent stems from an error about
the self, whereas the miscalibration of
the highly competent stems from an
error about others [2].
The first half of this is easy to see:
People who are new at something do
not know how much more there is to
learn, so they often rate themselves as

competent. Thats beneficial, as it lets
us keep learning new things without
initial discouragement!
But the second half is the part that
stalls our learning and progress. Once
you start to gain enough skill to judge
what a terrific skill would look like, you
realize how far away that might be. You
then see others who are considerably
more skilled than you are, and discouragement can set in.
Everyone makes a mistake here:
comparing without context If I compare
my skill at Java to another engineers, I
dont know what it took for them to get
to their current skill. I dont know what
other efforts or distractions they had.
The fix here is simple. In the first step
of this approach, you started logging
your work and successes. That gives you
Image by Valentina Photo / Shutterstock.com
mpostor syndrome is defined as

a psychological phenomenon in
which people are unable to internalize their accomplishments [1].
It triggers thoughts like: I dont belong; someone made a mistake when
they hired me. Or, My ID badge didnt
work on the door. Oh my God! They
caught me! Im fired.
This affects people across all races,
all genders, and all ages. It most commonly affects those who are new to
their work, and those who dont have
several role models in their workplace.
It leads to the burnout of strongly capable people. Left unaddressed, it reduces what people can do in their personal
and professional lives.
Impostor syndrome affects most
people at some point during their careers. Online estimates put the number at 70 percent of engineers, while a
quick poll of 30 personal engineering
contacts put it above 90 percent. The
first step to fighting it is easy enough:
Simply acknowledge impostor syndrome as a common feeling, built into
human psychology. If that doesnt
work, here is a five-step approach to
prevent it from slowing you down.
When a hospital employs more empathetic

nurses, its ratio of positive to negative sentiments
on social media jumps from 2.4:1 to 3.3:1.
one person you can compare against,

with full context, and in a healthy way.
You compare and compete against
yourself. This helps stop comparisons
against others, which is a common recurring trigger for impostor syndrome.
STRATEGY #3: SET GOALS

When you compare against yourself it
helps to have set goals, so you can see
how youre doing and what you set out
to do. It is important to never lower
your self expectations based on impostor syndrome. Set your goals high, even
if you miss them from time to time.
You might set goals yearly, quarterly, or even weekly. They should be
easy to grade. Examples include: learn
enough ObjectiveC to show a demonstration app, learn enough French
to order dinner, or get an A in data
structures.
Each person and each project at
Google has quarterly goals, which they
set themselves. At the end of each quarter, they score the results, and partial
success counts. Getting three quarters
of the way there scores as a 75 percent
for that goal. If an engineer always hits
all of their goals (averages 100 percent),
they know theyre aiming too low. If
they get half of the goals or less, they
know they set the goals too high. An ideal success rate is 70-80 percent, which
means successful engineers miss 20-30
percent of their goals, every quarter.
Setting goals like this is healthy,
while still pushing strong performance. Failure after strong effort is a
part of trying hard, and learning to fail
is an important skill to build.
STRATEGY #4:
AVOID SELF SABOTAGE
Your tracking needs to be honest. If
something fails, do not be negative, but
do not excuse yourself from accountability. Do not dwell on failures, but do

a quick retrospective analysis to help
avoid the same failure going forward.
The postmortem is a technique
used at Google to learn from failure.
The postmortem is a concise written
document that explains the following:
1. What went wrong? Add enough
detail to bring it back to mind later.
2. The cost of it going wrong. How
bad was it?
3. A triage list. What tasks do you
have to do to fix it?
4. A preventative list. What can you
change so this will not happen again?
There is no punishment or negativity in a postmortem. Punishing ourselves for failures does not help most
people learn to succeed.
Google postmortems are public to
any employee who could potentially
repeat the mistake. For use in personal
tracking, the document can remain
private.
Failures have costs, which can include time, money, reputation, and/or
opportunities. Since we have already
paid the cost, a postmortem helps to
reinforce the lessons learned, which
can make the costs worthwhile. Make
the postmortems preventative list part
of your future goals.
other people around you will provide

ongoing support.
If you get tired or stuck trying to work
on this, actively find help. Ask someone
you respect and trust (a professional, a
mentor, a leader in your community, or
a family member who knows you well).
These people can offer perspective and
tools for calibrating your judgment and
for staying engaged.
YOU ARE NOT ALONE

By keeping track of our successes,
comparing against ourselves and not
others, setting aggressive goals in a
reasonable way, avoiding self sabotage,
and seeking additional help, we can effectively and mitigate the harmful effects of impostor syndrome. Thus we
can improve both our happiness and
productivity in our careers and everyday life.
Because enough people are reporting impostor syndrome, it is a regular
topic at colleges, corporations, and
conferences worldwide. However, it is
something that you can work on and
work through. Give it some thought,
and discuss it with a good friend or a
friendly peer; you may be surprised at
their response.
References
[1]
STRATEGY #5: FOLLOW UP

AND GET HELP
There are a lot of talks, papers, and other resources about impostor syndrome
available online; a quick web search
will turn up quite a bit. Go do some research, as unblocking your success is
worth direct effort on your part.
Impostor syndrome is definitely
something to discuss with friends and
peers. It does not have to be a deep
chat, but just confirming it affects
Impostor Syndrome. Wikipedia, The Free

Encyclopedia. Wikimedia Foundation, Inc. 27 Oct.
2014. Web. 28 Oct. 2014.
[2] Kruger, J., and Dunning, D. Unskilled and unaware

of it: How difficulties in recognizing ones own
incompetence lead to inflated selfa ssessments.
Journal of Personality and Social Psychology 77,6
(1999), 11213 4.
Biographies
Dean Jacksons a member of the ACM and a technical
program manager working at Google, focused on Search
and Security.
Taliver Heath works in the technical infrastructure area
of Google at the Mountain View campus, and has been
facilitating the Life of an Engineer program to new hires
for the past five years.
13
begin
The Google-founded Calico lab researches age-related

illnesses, such as Parkinsons, ALS, and depression, which
together affect more than 40,530,000 U.S. citizens.
BLOGS
My $300 Home Cloud

Server: A Story of Blood,
Sweat, and eBay
By Wolfgang Richter
I really wish I had a dedicated Linux computer to run computer vision algorithms on, said my fiance a couple of
weeks ago. If you were there you would have been blinded
by the metaphorical light bulb that lit over my head. You
see, just the week before, my friend and co-worker had ordered an old, decommissioned (complete with non-classified stickers!) Apple Xserve off of eBay for merely $40.
Like my fiance, he wanted to have a machine for a special
purpose: test compilations of open source software on a
big-endian architecture. I was quite envious that he was
able to hack on such cool hardware for such a cheap price.
But, I wasnt yet ready to bring out my wallet. I couldnt
14
justify indulging a new hobby without good reasonI was

stuck waiting for just the right impetus. I didnt wait long.
My fiances wish became my command!
I immediately related the story of my co-worker to her:
A server for just $40! Granted, we would want a littleendian, x86-64 architecture. Plus, for her algorithms
and my virtual machine research wed probably want a
lot of cores and as much RAM as possible. Oh yeah, did
I mention I also wanted a beefier machine at home so I
could manipulate large virtual machine images? Virtual
machines (VMs)! Wed need CPUs with VT-x or AMD-V so
we could run VMs with accelerated hardware support.
VMs run slow as a snail without acceleration. That would
make the machine useless to me.
Focused on my quest, I started scanning eBay listings
daily. My co-workers even began to notice and started
asking me if I was looking for something specific.
I responded I was toying with the idea of trying to
grab some cheap data-center-class hardware (for the
astute, cheap data-center-class hardware should be
Apple has been working on wearable health sensors

for example, apps that detect heart attacks before they
happensince the original iPhones debut in 2007.
an oxymoron). I was worried my project would end in

failure, and wasnt quite ready to announce to the whole
world my larger plans. After several days of failed bid
attemptsI always seemed to get sniped in the last few
secondsI finally found what appeared to be the perfect fit.
There is a lot of conjecture on where the Dell 1U
rackmount model CS24-SC came from [1]. Some people
say Facebook data centers [2]. Others just say that it
was mass-produced for clouds [3]. Whatever these
servers were used for, they were all retired by the
thousands and can now be found all over eBay. The
general consensus is Dell never sold these to general
customers; the CS24-SC was a special custom-designed
server sold by the tens of thousands to certain large
customers. Thus, the CS24-SC has no support from Dell
[4]. I havent been able to find anything outside of what
random other CS24-SC owners have found in the years
since the great decommission event.
The CS24-SC has a few variations, but they dont
deviate too significantly. The one I had in my sights
came with two quad-core Xeon E5410 @ 2.33 GHz CPUs.
OK, fairly beefy compute from a few years ago giving
us eight total real cores. It had 8 GB of RAM installed,
which felt a little wimpy. Articles from second-hand
owners online were conflicted on the maximum amount
of RAM supported by the CS24-SC. Some said 48 GB [3],
others said 24 GB [2] was the max. Well, it didnt matter,
because the greatest amount of RAM I could find at a
reasonable price was 24 GB of data-center-class ECC RAM
(PC2-5300P for the interested). Cool, what were we still
missing? Oh, most of these second-hand machines dont
come with hard drives or hard drive caddies.
After a quick visit to Newegg, I identified a cheap
Seagate 1 TB hard drive (ST1000DM003) to slot in. Meh,
lets do this right and add in an extra hard drive for
RAID1 to protect our work. I also threw in two CAT6
Ethernet cables so we could use both of CS24-SCs
gigabit network ports, and a power cable. Well, thats
about it right? The server on eBay had its own case,
400-watt stock power supply, motherboard, and other
needed components.
We had to wait 1.5 weeks, but finally the CS24-SC
arrived. I anxiously picked it up at our local FedEx
location just up the street. My fiance and I unboxed it
together and hooked up all the components together.
She tried putting in some of the RAM herself, so this
counts as a date right? We were both worried that it
wouldnt boot, and in a sense that became a self-fulfilled
fear. After plugging in a VGA monitor, we just had a
Figure 1. NewEgg shipped these within a day!
Figure 2. Dell CS24-SC sticker, its authentic!
15
begin
Figure 3. Its alive!
Figure 6. See, it really does work.
Figure 7. Always check for bad hardware.
Figure 4. The HDDs connect fine without caddies.
Figure 8. Installing Ubuntu.
Figure 5. Cabling in the back.
16
400,000
The approximate number of people worldwide who have received

Cochlear implants, a device that bypasses damaged inner-ear
organs and directly stimulates the auditory nerve.
black, blank screen. Uh oh, maybe the hardware is bad?

Or maybe the RAM is bad?
I really racked my brain thinking of ways to check on
this system. I plugged in the Baseboard Management
Console (BMC) port into my router. Based on its
DHCP client table, I guessed a certain device on my
network was coming from the BMC port. My hunch was
confirmed when I port-scanned and discovered port
81 open and running an Apache server. After going to
the server in my browser I was presented with a login
prompt. I was getting desperate and worried. I thought
that even if the VGA port was bad for some reason, wed
at least be able to get into the remote console. But how
to get past this login screen?
I tried several username/password combinations, and
luckily root/root worked. I found out later online that is
the default username/password combination [5]. Thank
you to whomever left this at the default, or reset it! If you
sell a server with such a management console, please
reset it if you customized it at all. It turned out the VGA
port wasnt bad, we just didnt have the monitor plugged
in before the BIOS flashed its screen. The system went to a
blank screen after failing to boot an OS.
Okay, phew, things seemed to be working. I downloaded
Ubuntu 14.04 LTS Server and copied it onto a USB stick.
Our CS24-SC had no trouble booting into the Ubuntu
installer off of USB. We installed Ubuntu, named our
server phoenix, after the ever-reincarnating mythical
bird, and started customizing our CS24-SC. The two hard
drives, 24 GB RAM, and whole system were recognized
perfectly by the BIOS and Ubuntu. The only lingering
issue I have is that Ubuntu doesnt seem to properly
display through the VGA interface after it boots. Grub
displays fine, and so do the early-stage kernel messages.
Perhaps this is just a driver issue I need to track down.
Also, the fans on the PSU dont spin, but it doesnt appear
to be going bad yet.
Hardware virtualization seems to work, and we are
setting up our own work environments within Vagrantmanaged VMs. Im using this opportunity to experiment
with some advanced Linux functionality Ive never tried
before. Our two hard drives are not a traditional RAID1.
Im using the new btrfs file system to mirror our root
partitions. There would be some work involved in setting
up the second hard drive to boot, but we wont lose our
data. I setup the dual gigabit ports into a single bonded
virtual device using the Linux kernels balance-alb
algorithm to try and balance inbound and outbound TCP
flows across both ports.
What were the total costs?

COMPONENT
QUANTITY
COST
SUBTOTAL
Dell CS24-SC Rackmount Server
$120.00
$120.00
24 GB PC2-5300P RAM
$64.98
$64.98
Seagate 1 TB HDD
$54.99
$109.98
CAT6 Ethernet
$1.99
$3.98
BYTECC Power Cable
$4.99
$4.99
Shipping
$2.99
$2.99
Taxes
$3.90
$3.90
Total
$310.82
There you have it, a[n] [old] data-center-worthy

home cloud server for only $310.28. Just for fun I tried
customizing a hypothetical order on Dells website for
new hardware configured the same way our CS24-CS is:
It came out to more than $3,300with discounts it only
drops to $2,400. Simply adding a second CPU on the Dell
website costs more than $500, more than our entire setup!
The more modern hardware is faster, but our little CS24CS is almost 90 percent cheaper. Thanks for reading my
story of how I built a $300 home cloud server. And now
maybe you can too with a little elbow grease and eBay.
tl;dr Quick Server Specifications:
TYPE
COMPONENT
CPU
Two Quad-Core Intel Xeon E5410 @ 2.33 GHz
Memory
24 GB ECC RAM
Disk
Two 1 TB HDDs (RAID1)
Network
Two Gigabit Ethernet Ports
References
[1] Dell CS24-SC Server. http://www.tedunangst.com/flak/post/Dell-CS24-SC-server.
[2] The Definitive Guide to the Dell CS24-SC Server: Drivers, Config & Tips. Hurtig
Technologies. June 5, 2014. https://hurtigtechnologies.com/2014/06/thedefinitive-guide-to-the-dell-cs24-sc-server.
[3] On the Dell CS24-SC Server. Rambling Geek. Nov. 13, 2012. http://www.
aramblinggeek.com/on-the-dell-cs24-sc-server.
[4] Dell CS24-SC Drivers. Dell.com Cloud Services Forum. 2012. http://en.community.
dell.com/support-forums/cloud/f/4715/t/19456940.
[5] Willis, R. Dell CS24-SC BIOS & BMC v2.5 Firmware Download. May 5, 2014. http://
robwillis.info/2014/05/dell-cs24-sc-bios-bmc-v2-5-firmware-download.
Biography
Wolfgang Richter is a fifth year Ph.D. student in Carnegie Mellon Universitys Computer
Science Department. His research focus is in distributed systems and he works under
Mahadev Satyanarayanan. His current research thread is in developing technologies
leading to introspecting clouds. tl;dr: Cloud Computing Researcher.
17
feature
Gathering People
to Gather Data
An interview with Paul Wicks, Vice President of Innovation at PatientsLikeMe,

a patient network and real-time research platform.
DOI: 10.1145/2676566
atientsLikeMe (www.patientslikeme.com) is a patient network and a real-time

research platform. Through the network, patients connect with others who have
the same disease or condition and track and share their own experiences. In the
process, they generate data about the real-world nature of disease, which helps
researchers, pharmaceutical companies, regulators, providers, and nonprofits develop
more effective products, services, and care.
This sharing of online medical data

has led to more than 50 novel studies
including: a patient-led observational
trial of lithium in ALS (amyotrophic
lateral sclerosis), new patient-reported
outcome measures in neurology, a
dose-response curve for the benefits
of friendship between epilepsy patients, and new methods for gaining
patient input into clinical trial design.
Specializing in the conduct of online clinical research, Paul Wicks
who is the Vice President of Innovationis responsible for shaping the
scientific validity of the PatientsLikeMe platform and generating insights from the personal health data
shared by members. In this interview,
Wicks illustrates the progress made
by PatientsLikeMe and new challenges facing the organization.
BACKGROUND
DIANA LYNN MACLEAN: Paul, you
are currently VP of Innovation at
PatientsLikeMe. Can you tell us
a little bit about what this means
18
and what drove you to pursue

what you do now?
PAUL WICKS: My job at PatientsLikeMe
boils down to this: Doing things for the
first time. Ive been lucky to be a part
of the team here for eight years now,
first as a community moderator, then a
scientist, and then leader of our research
and development team. During that
time weve expanded from a handful of
neurological conditions like amyotrophic
lateral sclerosis (ALS), multiple sclerosis
(MS), and Parkinsons disease to more
than 2,000 health conditionsmeaning
that anyone can join. The main core of
the platform involves patient-reported
data on conditions, treatments, and
symptoms. While theres no better source
for how a condition affects people than
patients themselves, there are some
issues with patient-reported information
that we are working through. These
include ensuring the data is kept up-todate, is accurately recorded, and that it
isnt affected by psychological factors.
The types of projects were looking to
develop are those that start to address
these issues. For example, weve worked

with Dr. Max Little to prototype tools that
gather voice data from a Parkinsons
patient from a phone call. Dr. Little
believes a 30-second callrecording
the signature of their voiceprintcould
be a more accurate, rapid, and objective
way of measuring their disease than
existing methods. Were using our Open
Research Exchange to help researchers
crowdsource the development of new
patient-reported outcome measures.
Were also charting the use of wearable
health sensors that passively and
continuously monitor activity and vital
signs. Perhaps the greatest opportunity
is in connecting patients to their medical
The Internet allows

you to reach patients
faster, cheaper, and
in a way thats more
convenient for users.
records so they can import the most

valuable data that they want to share with
other patients or for research.
I originally got into this area back in
2002 through a Ph.D. [program] studying
ALS at the Institute of Psychiatry in
London. By day I would go out and visit
patients to give them a battery of
cognitive tests; by night I would moderate
an online forum for people with the
disease to connect to one another. The
Internet was nowhere near as ubiquitous
then, but the opportunity seemed
ripewhy see one patient a day when
you could gather data from hundreds
or even thousands all in one go? As a
trainee scientist, it felt like most of my
time was spent doing the logistics of data
collection rather than actually thinking
of new hypotheses or analyzing data.
If we could get vast amounts of data
continuously, then we should be able to
dedicate more of our thoughts to what
were meant to be doingunraveling
diseases! In 2006 some of the users of
the forum I was running became the beta
testers for PatientsLikeMe, which was
founded by a family affected by ALS over
in Cambridge, Massachusetts. As soon as
I saw the site I knew they had something
really specialthe same research tools
Id been administering were right there
for patients to use. Even better, they
provided visualizations to help patients
understand their condition in context,
and connect with other patients who had
walked in their shoes.
Although I trained in quite a traditional
environment relative to the world of
startups and technology, Ive always
tried to keep one foot grounded in
academic medicine. One of the first
things we did when I started working
with PatientsLikeMe was to try and get
our first peer-reviewed scientific paper
published. During my Ph.D., I had come
across a handful of patients who yawned
uncontrollably, sometimes hundreds
of times a day, so much that their jaws
dislocated. But as they say, the plural
of anecdote is not data, so I could
never get anywhere having only studied
about 90 patients in my research. When
I stumbled across a paper reporting
uncontrolled yawning as a side effect of
anti-depressants, I saw my chance. We
added excessive yawning as a symptom
in our ALS community and within weeks
had data from 254 ALS patients, a
dataset that would have taken me five

years to gather through traditional
methods given the rarity of the disease.
Not only did we find excessive yawning
was relatively common (occurring as a
severe problem in about 9 percent of
patients), it was more common in certain
subsets of the disease affecting the
mouth and throat, although there was no
correlation with measures of breathing
severity. We hadnt just identified an
unmet need among patients; we probed
the underlying pathology of the condition,
and had gone from concept to publication
in a couple of months. From that point on,
I was hooked.
TECHNOLOGY, PATIENTS
AND HEALTHCARE
DLM: The practice of patients sharing
health information with each other
online manifested as soon as the
Internet became publicly available
in the mid-90s. How has sharing
health information online changed the
landscape of medicine and healthcare
from the perspective of both patients
and healthcare professionals?
PW: I think weve seen a shift in the
past decade or so from those early textdriven set of interactions, which were
anecdotal stories from text-based forums
and message boards, to a more data-
19
feature
driven approach where patients want to
help develop new measures, improve the
clinical trial protocols, read the peerreviewed papers for themselves, and even
run their own studies. For both patients
and healthcare providers we see huge
variation between conditions, and there
are all sorts of different factors that go
into that. The parents of a child with a rare
developmental condition want to learn as
much as they can, and will rapidly get to
the stage where they know as much as the
small number of experts out there (which
arguably isnt much). In cancer the state
of medicine has advanced to the point
where youve got very deep molecular
genetics, personalized treatment
regimens, imaging, all of which are difficult
for a patient to get a hold of and which
are changing all the time. An ALS patient
might be willing to self-experiment
because they might feel they have few
other options, whereas a psoriasis patient
whose condition flares up every few
months might be less willing to take risks
with experimental treatments.
We see the same level of variability
on the health professional side. For
instance in ALS, our most developed
condition where we have a decade of
experience, we work with a consortium
of more than 80 clinical experts called
ALSUntangled, who use the Internet
to engage with patients who want
experts to investigate complementary
and alternative medicines. The group
reads the scientific literature, checks
their case files, and reviews data from
PatientsLikeMe, then publishes their
findings open access in the main ALS
journal [http://informahealthcare.
com/loi/aml]. There, you can really see
healthcare professionals committed
to meeting patients where they are,
and responding to their questions
respectfully. I think healthcare providers
are pleased when their patients take
a more engaged and activated role in
managing their own condition, but a few
can sometimes feel threatened when
patients begin educating themselves.
In fact, PatientsLikeMe published
data suggesting about 10 percent of
PatientsLikeMe users change physicians
as a result of information theyve learned
from the site.
I think many healthcare professionals
are interested in the potential of all
these technologies, whether theyre
20
digital communities or apps or wearable

devices, but the vast majority of these lie
outside the traditional healthcare system.
Professionals are under pressure from all
sides, including ensuring they are getting
reimbursed for their time. If the fee
for service world were instead a world
in which fees were paid for improved
outcomes, it would likely be easier for
these professionals to engage with
innovators.
EFFICACY OF ONLINE
HEALTH COMMUNITIES
DLM: What are some of the benefits
that patients derive from participating
in online health communities?
Correspondingly, what are some of
the pitfalls, and what technological
advances (form factors, interface design
techniques, algorithms, etc.), if any, do
you see addressing these in the future?
PW: Weve published several surveys
reporting the benefits PatientsLikeMe
members experience, such as learning
about a new symptom (72 percent
of survey respondents agreed),
understanding the side effects of a
treatment (57 percent agreed), and
finding another patient who had taken a
treatment they were taking (42 percent
agreed). Weve even seen benefits that
might be clinically relevant like improved
medication adherence in HIV, reduced
self-harm in mood disorders, and greater
seizure control in epilepsy. Intriguingly,
in a follow-up study in our epilepsy
community, we found the greatest
predictor of benefits experienced by
patients was the number of friends with
their condition they had made on the site.
As it turns out, patients themselves are
Many healthcare
professionals are
interested in the
potential of all these
technologies, but
the vast majority
of these lie outside
the traditional
healthcare system.
the active ingredient in intervention.

Were just finishing up a report of a study
recently completed with the Department
of Veteran Affairs in epilepsy, where we
recruited veterans with epilepsy to see
how PatientsLikeMe might affect their
self-management and self-efficacy. The
results should be published shortly, but
its been heartening to see that other
groups can detect these benefits too. Its
not just us tooting our own horn!
The major pitfall I see with nearly any
health technology is theyre a clunky bolton to your daily life. When youre feeling
better you dont want to be reminded of
your illness and devote time to entering
data, slapping on a wearable device, or
filling out a survey. You just want to get
on with living. Thats why we need to learn
how to harness the whole tapestry of big
data being collected, from smartphones
to pharmacy loyalty cardsall with the
appropriate permissions, of course
with an emphasis on returning value
to patients, not creating commercial
opportunities for exploitation.
THE VALUE OF
PATIENT-GENERATED DATA
DLM: As more people turn online for
medical advice, the quantity of medically
relevant, patient-contributed data
available online continues to grow.
What is the inherent value of this data,
and could you give an example of an
interesting discovery or insight that
derives from patient-contributed data?
PW: The Internet allows you to
reach patients faster, cheaper, and in a
way thats more convenient for users,
and lowers the barriers to conducting
research. Of course youve still got to
design your study well and realize your
audience is easily distracted: Youre
competing for their attention with
Facebook and Candy Crush Saga.
There are few different classes
of patient data we see online. One is
just a straightforward like-for-like
of traditional methods for capturing
data such as postal questionnaires,
interviews, or telephone surveys. The
second type of data is information that
used to exist in silos but is now being
shared openly. So, for instance, patientreported outcomes (PROs) had been used
in clinical trials for 30 years, but it was
information taken from patients and then
locked away. Now, were seeing patients
who can enter their PRO data, use it

as a tool for gathering insights about
themselves, communicate with other
patients, and contribute data that can be
aggregated by multiple researchers.
The third type of data is entirely new.
It includes the semantic density of email
messages, sensor data from the GPS in
smartphones, and biometric data from
wearable devices. These things would
have been high-end R&D tools in a few
highly selected places on Earth just 10 or
20 years ago. Now, theyre at everyones
fingertips. There are studies and tools
that can harness all three types of data
and mash them up in new and powerful
ways. Thats why Im so excited to get up
for work every day.
DLM: What are some of the main
challenges in extracting useful insights
from these data, and how do you see us
addressing these in the future?
PW: Weve been making incremental
improvements since the beginning.
Today PatientsLikeMe gets an upgrade
about every two weeks. However to really
catalyze this whole field, I think we need
to look at the consumers of this data,
the decision makers, and ask them what
would it take for you to integrate patient
data gathered online into the decisions
you make every day? In healthcare
those are decisions about what drugs to
approve, the value at which treatments
should be reimbursed, the definition of
good quality carebig decisions. When
those decision-makers signal that yes,
under the right circumstances they would
take patient-generated data seriously,
then youve just aligned a whole group of
individual stakeholdersentrepreneurs,
academics, patients, investors,
advocatesto solve the challenges in
front of us.
The big challenges are the usual
suspects: data quality, data density,
validation, reliability, bias, and
generalizability. But the good news is
were working on these issues every day.
And because this whole ecosystem is built
upon technology, were going to benefit
from consumer technologies like tablets
or smartwatches as they become more
widespread too.
CHANGING MEDICAL RESEARCH

DLM: Some of your recent work has
explored the idea of using online
The fact that they

have to share their
data online is an
indictment of the
healthcare system
why cant they access
their medical records
online in a way thats
meaningful and
valuable to them?
communities to drive, design, and even

run clinical trials. Could you describe how
this would work, and what it would take
to meet standards of medical validity?
PW: When we started operating as
a business, one of the ways we made
money was by raising awareness about
specific clinical trials. In one example
we helped Gilenya recruit MS patients
for their clinical trial, and they stated
publicly that we sped up recruitment by
several months. Later, we built a mashup with ClinicalTrials.gov so any patient
could find out about any clinical trial for
which they might be eligible, regardless of
whether it was sponsored or not. But we
knew just feeding the existing machine
wasnt enough, so we sought to find ways
patients could really change the system.
In many ways, we followed the lead of
patients themselves rebelling against the
system and breaking the norms of clinical
trials. Back in 2008 a small Italian study
was published. This study suggested
lithium carbonate effectively delayed
the progression of the normally fatal
disease ALS. Although that study was very
small, just 16 treated patients and 28
controls, it caused a big stir in the patient
community and within six months we
had 10 times that number. One hundred
and sixty patients, who had got hold of
lithium off-label, were sharing their data
on PatientsLikeMe to see if the drug was
working. Some clinicians thought this was
dangerous and might harm recruitment
for the formal randomized control trials
to come, but patients werent satisfied to
sit on the sidelines when an opportunity
to halt their disease might be within

reach. Given the slow pace of medical
research, most of the patients who opted
to take lithium would have been ineligible
for the carefully controlled trials that
would take years to enroll, only to have a
50/50 chance of receiving a placebo and
being kept in the dark until the study had
finished. We partnered with some of the
pioneering patients in that community to
upgrade our tools to allow the recording
of dosage levels, side effects, and,
crucially, the same patient-reported
outcome measures widely used in clinical
trials. By visualizing the data continuously
and in real time, we quickly got the
sense that lithium was not a miracle
drug, and when we subjected the data to
more rigorous analysis, using data from
hundreds of historical matched controls,
we were able to say more conclusively
that lithium didnt work. This finding was
subsequently replicated by a number of
clinical trials that followed in years to
come. In terms of credibility and validity,
one thing weve learned is when youre
doing something new youll be held to
even higher standards than the status
quo. Thats why we published our findings
in a high impact peer-reviewed journal
Nature Biotechnologypublished all the
details of our matching algorithm, made
sure the paper was open access so that
anyone could read it, and even uploaded
a de-identified copy of our dataset so
others could replicate our findings. Id like
to say this made our position unassailable
but, even today, the ALS field continues
to minimize the impact of this study and
claims because it was patient-reported
the finding was null and void.
Weve also taken another approach
to clinical trials, which is to, rather than
condemning the whole system, ask
what can patients do to optimize it to
be more patient-centric. This year weve
undertaken a number of projects with
commercial sponsors to gain patient
feedback on aspects of the trial protocol,
recruitment materials, and outcome
measures so that trial designers can
make trials more appealing to patients,
less burdensome, and ultimately more
aligned to what matters to patients. In
this way we hope we can bridge a gulf
that risks developing between trial
designers focused on their scientific
and commercial goals versus patients
trying to improve their outcomes. Its still
21
feature
early, and only a tiny handful of trials are
designed with systematically gathered
and robust data about patients views,
but I hope we can help to change that.
Unlike the scientific community we find
businesses can be more responsive.
If we can show we help avoid a costly
commercial failure or an expensive
protocol deviation that slows down their
path to regulatory approval, then theyre
willing to accept evidence that it works.
TACKLING A VARIETY
OF CONDITIONS
DLM: Are certain medical conditions
more amenable to being the focus of a
successful online health community?
If so, are there particular attributes
(rarity/prevalence, stigma, etc.) that
correlate with this?
PW: Absolutely, we even have
an internal term for it; we call it the
PLMability of a condition. Where we
seem to do best is in chronic conditions
that have a major impact on patients
lives, where they feel directly affected
by the condition (unlike, say, high blood
pressure), and where their own knowledge
and behavior is likely to have an impact
on their outcomes. For instance we often
say if you break your leg in an accident,
you wouldnt necessarily need a system
like PatientsLikeMe. You have an acute,
very curable condition that will be getting
better soon, and the medical system
knows how to treat it. Contrast that with
a neurological condition, like multiple
sclerosis, where youre going to live with
it for the rest of your life, we really dont
know what causes it or how to treat its
many varied symptoms, and your needs
may change dramatically over time.
Even within those conditions for which
were highly suitable, we see variations in
when patients come to us in the course
of their disease. We see a lot of people
whove been recently diagnosed who are
just trying to find out everything they
can, and we see another group of relative
veterans who have been managing their
condition well for years but are now
experiencing something new they dont
know how to deal with. They come back to
the community to draw upon the wisdom
of the crowd.
Sometimes theres a conflict between
the research needs of a scientist and
the lived experience of a patient. For
example we have a community of organ
22
We need to learn
how to harness the
whole tapestry of big
data being collected
with an emphasis
on returning value
to patients, not
creating commercial
opportunities for
exploitation.
transplant patients, and when theyre on

the waiting list or awaiting surgery then
their engagement is very hightheyre
looking for tips, sharing information, and
looking for support. But once theyve had
the transplant, assuming its successful,
then they just want to get on with the rest
of their lives. That means we do probably
have unknowable biases in terms of who
keeps coming back, and again thats why
were interested in exploring other ways of
receiving information from patients that
might be useful to them if they need it,
but that doesnt require they keep coming
back to submit more data.
PRIVACY
DLM: Data privacy is a growing concern
in todays technological landscape, and
legislation around safeguarding medical
information is particularly strong. Yet
patients seem willing to share detailed
health information with each other in
public online forums. Why do you think
this is, and how do you see it playing out?
PW: There are a few factors at play
here. On our site, patients are making
an informed decision about the risks
and benefits that sharing their health
data online might bring to them. (See
our privacy policy here: http://www.
patientslikeme.com/about/privacy.)
Theyre only sharing as much information
as they feel comfortable with. For
instance if theyre on the site reporting
data about lung cancer, they might not
be reporting another condition they have
such as a mental health diagnosis, and
thats fine. Theyre putting in and getting
out what they want to. By offering their
experience they are vulnerable, but they

also realize they are among friends and
hopefully they benefit enormously from
what others like them have shared in a
vulnerable moment, so the virtuous cycle
of sharing data becomes apparent.
In part, the fact that they have to
share their data online is an indictment
of the healthcare systemwhy cant
they access their medical records online
in a way thats meaningful and valuable
to them like their personal health record
is on PatientsLikeMe? When you look at
the emphasis of most health IT systems
though its billing, locking the data down in
silos, and driving lower cost care through
physician workflows. Im afraid to say
patient benefit is either low on the priority
list of most electronic-record systems or
absent altogether.
Ive been doing this for eight years
now and helped to build communities
of more than 250,000 patients and
Im pleased to say I have yet to hear
about a major breach of medical privacy
resulting from sharing data online, but
its certainly a concern. Then again
when online banking first appeared, the
world seemed terrified of identify theft,
and while thats certainly been a valid
concern its also the case that we went
ahead and rolled out online banking to
the whole world to great benefit.
Privacy is certainly a very real
concern, but I think its time we gave
more credit to patients in making their
own decisionspeople can drink or
smoke or drive cars or go skydiving or
buy a handgun, but for some reason when
they become ill or want to share medical
information about themselves, the health
system suddenly become very concerned
that they dont know how to act in
their own best interest. Its certainly a
rapidly evolving area and we are always
conscious of the trust patients put in
us with their health data. Its up to us
to honor that trust and always try to be
one step ahead of those that would take
advantage of their generosity.
Biography
Diana Lynn MacLean is a Ph.D. candidate in the Computer
Science Department at Stanford University, where she is
advised by Jeffrey Heer. Her thesis comprises crossdisciplinary research focused on extracting medically
relevant data from patient-authored text on the topic of
substance abuse. Out of the office, Diana is an avid rock
climber and Tomb Raider fan. Diana received her B.A. from
Harvard University in 2009 in computer science.
2014 ACM 1528-4972/14/12 $15.00
Opportunities of
Social Media in Health
and Well-Being
Intelligently leveraging data from millions of social media posts is a modern
public health approach that has the potential to save many lives.
By Munmun De Choudhury
DOI: 10.1145/2676570
eople are increasingly using social media platforms, such as Twitter and Facebook,
to share their thoughts and opinions with their contacts. One in six people in
the world today is a user of Facebook [1]. In a way, social media has transformed
traditional methods of communication by allowing instantaneous and interactive
sharing of information created and controlled by individuals, groups, and organizations.
An important attribute of social media is that postings on these sites are made in a
naturalistic setting and in the course of daily activities and happenings. As such, social
media provides a means for capturing behavioral attributes that are relevant to an
individuals thinking, mood, communication, activities, and socialization. Moreover,
this real-time data stream of social information is often annotated with context including location information,
cues about ones social environment,
and rich collections of multimodal information beyond text, such as images
and videos.
With the increasing uptake of
social sites, there has been a corresponding surge of interest in utilizing
continuing streams of evidence from
social media on posting activity to reflect on peoples psyches and social
milieus. In fact, the ubiquitous use
of social media, as well as the abundance and growing repository of such
data, has been found to provide a new
type of lens for inferring healthrelated behaviors and mechanisms
[24]. A common thread in this body
of research is how computational

techniques may be applied to naturalistic data that people share on todays
online platforms in order to make
sense of their health behaviors and
related experiences. In the context of
this surge in research interest, this
article highlights the potential opportunities and challenges in the use of
social media as a novel stream of information for augmenting traditional
approaches of health assessment.
ROBUST, LARGE-SCALE SOCIAL

MEDIA MARKERS OF HEALTH STATES
Leveraging Internet data for modeling and analyzing health behaviors
has been a ripe area of research in the
recent past. Google Flu Trends (http://
www.google.org/flutrends/) provides
nuanced predictions of flu infections

based on online search queries. Paul
and Dredze developed a disease-specific topic model based on Twitters posts
in order to model behavior around a variety of diseases of importance in public health [5]. Through language modeling of Twitter posts, Culotta found
evidence of high correlation between
social media signals and diagnostic
influenza case data [6]. Sadelik et al.
developed statistical models that predicted infectious disease spread in individuals based on geotagged postings
made on Twitter [4] (also see Moreno et
al. [7]).
In the behavioral health domain,
Park et al. found initial evidence that
people do post about their depression
and even their treatment for depres23
feature
24
NEW ASSESSMENT TOOLS

Due to the richness of data generated
on online social platforms, we are provided new opportunities for understanding health behaviors of individuals, in a scope and scale not possible
before. For instance, surveys and wearable sensing tools that are often used
extensively for personal health monitoring can indicate ones geographic
location, as well as other physiological
and affective responses [10, 11]. However, they generally cannot capture the
context and content of these reactions.
Conversely, SNSs provide a rich
ecosystem where the context and
content of ones affective, behavioral,
and cognitive reactions, as well as
social interactions, can be observed
over extended periods of time [9]. The
characteristics of such context and
content can be learned for thousands
or even millions of people. These social factors are known to be key in
the detection and assessment of a
number of health conditions and outcomes, and can be made to work in
a complementary fashion alongside
traditional approaches.
Moreover, in typical behavioral surveys, responses are prompted by the
experimenter and typically comprise

recollection of (sometimes subjective)
health facts. Social media measurement of behavior captures social activity and language expression in a naturalistic setting. Such activity is real
time, and happens in the course of a
persons day-to-day life. Hence it is less
vulnerable to memory bias or experimenter demand effects, and can help
track health concerns at a fine-grained
temporal scale.
NOVEL PLATFORMS OF
PSYCHOSOCIAL SUPPORT
Self-disclosure is an important therapeutic ingredient [12], and is linked to
improved physical and psychological
well-being. In fact, self-disclosure has
received a great deal of attention in
counseling research because of its hypothesized benefits for the client during the course of therapy, such as an increase in positive affect and a decrease
in distressing symptoms [13]. Jourard
reported the process of self-disclosure
was a basic element in the attainment
of improved mental health [12]. Ellis
reported discourse on emotionally laden traumatic experiences can be a safe
way of confronting mental illness [14].
Image by Everything Possible / Shutterstock.com
sion on Twitter [3]. Kotikalapudi et

al. analyzed patterns of web activity
of college students that could signal
emotional concerns [8]. In other related work, our past research examined
linguistic and emotional correlates
for the postnatal course of new mothers, and thereafter built a model to
predict extreme behavioral changes
in new mothers [9]. Our findings indicated behavioral concerns such as
post-partum depression may be reflected in mothers social media use:
including lowered positive affect and
raised negativity, and use of greater
first-person pronouns indicating
higher self-attentional focus. In fact,
the behavioral changes of mothers
could be predicted by leveraging their
activity from simply the prenatal period.
A common thread in this body of
work is that these methods, which intelligently leverage data from millions
of social networking site (SNS) posts
and users, have demonstrated efficiency in performance and accuracy
when applied to a number of health domains. Together, they point to the potential of social media as a signal to leverage in the study of health concerns.
On similar lines, seminal work by Pennebaker et al. found participants assigned to a trauma-writing condition
(where they wrote about a traumatic
and upsetting experience) showed immune system benefits [15] (see also
Ramirez-Esparza et al. [16]). Disclosure in this form has also been associated with reduced visits to medical
centers and psychological benefits in
the form of improved affective states
[17, 18]. Rodriguez similarly found revealing personal secrets to an accepting confidant could reduce the feeling
of alienation and, as a consequence,
can also lead to health benefits [19].
Social media platforms are known
to allow increased self-disclosure [20],
allowing individuals to discuss sensitive or otherwise considered stigmatic
health topics with communities they
identify with. Because users can be
essentially anonymous or pseudonymous on social media, and therefore
are not bothered by self-presentation
or concerns related to tracking their
history on the site, these services can
facilitate fruitful connections among
peers with similar stigmatic experiences and provide an open and honest
platform of discourse.
REACHING LARGE,
DIVERSE POPULATIONS
Social media and SNSs are being increasingly adopted across different
walks of life. According to a Pew Internet study [1], currently 73 percent
of online users use at least one online
social platform, with Facebook being
the most popular. Beyond the everyday
use of sharing details about the mundane goings on of life, 59 percent of
U.S. adults have used online resources
to obtain health information in the
past year. In the context of health
and well-being, social media use can
serve a range of purposes, including
Proactively using
social media to
increase public
awareness of and
education on health
issues is a logical
modern public
health approach.
seeking advice, connecting with experts and individuals with similar

experiences, sharing questions and
concerns around treatment options,
and understanding professional diagnoses. Online health content can
enhance coping and self-efficacy [21],
affect health-related decisions and
the behavior of users and their friends
and family, enable better management of chronic health conditions,
and fuel discussions with healthcare
providers [22].
Hence there is an opportunity for
healthcare and clinical professionals to leverage such online presence
to reach a variety of audiences and
provide health help and resources for
timely intervention and management
of a condition. For instance, women
tend to use Pinterest and Tumblr more
[1], while use of Instagram and Twitter is notably high among AfricanAmerican and Hispanic audiences.
Potentially, such affordances can be
nurtured to target public-health messaging for these demographic groups.
NEW OPPORTUNITIES FOR PUBLIC

HEALTH SERVICES
SNSs provide a unique avenue for
25
feature
ACM
ACM Conference
Conference
Proceedings
Proceedings
Now
via
Now Available
Available via
Print-on-Demand!
Print-on-Demand!
Did you know that you can
now order many popular
ACM conference proceedings
via print-on-demand?
Institutions, libraries and
individuals can choose
from more than 100 titles
on a continually updated
list through Amazon, Barnes
& Noble, Baker & Taylor,
Ingram and NACSCORP:
CHI, KDD, Multimedia,
SIGIR, SIGCOMM, SIGCSE,
SIGMOD/PODS,
and many more.
For available titles and

ordering info, visit:
librarians.acm.org/pod
26
public health organizations to reach

out, monitor, and support large and
diverse populations. Clinicians and
caregivers can tune in to social media
conversations in real time, listen and
collect feedback, identify information
gaps, and quell misconceptions about
health needs of individuals. Additionally, due to the multi-way, interactive
functionality that is inherent to these
platforms, social media can allow
these agencies to increase direct engagement to maintain and increase
trust and credibility about the variety
of health information that surfaces on
these platforms.
There is also potential for public
health organizations to be able to engage with opinion leaders and influencers on health-related topics in social media, and their conversations.
Influencers can be both organizations
and individuals and exhibit the characteristics of credibility, persistence
in convincing others, and ability to
drive conversations so that others take
notice of the topic or idea and show
support. Potentially, by engaging with
such influencers, public health officials can discuss ways to promote
messaging on shared communication
goals to increase the reach of public
health communications.
RISK TO VULNERABLE POPULATIONS

Despite these positive benefits in
health assessment, social media may
pose a hazard to vulnerable populations through the formation and influence of extreme communities
groups on SNSs that promote and
provide support for beliefs, attitudes,
and behaviors considered typically
harmful or unacceptable by the social
mainstream. Examples include proanorexic behavior, pro-suicide tendencies, deliberate amputation, or other
forms of self-harm. Essentially, for
these populations, social media may
provide them with an environment
and avenue to seek and provide support and acceptance that is difficult
to obtain through offline means. Although these online groups may provide the benefit of support, they may
present a risk to the public by encouraging vulnerable individuals to hurt
themselves. Additionally, because the
web eliminates geographic barriers
Due to the richness

of data generated
on online social
platforms, we
are provided new
opportunities for
understanding
health behaviors
of individuals.
to communication between people,

the emergence of pro-self-harm social
media sites and content may present
a new risk to vulnerable people who
might otherwise not have been exposed to these hazards.
PRIVACY AND ETHICS OF UTILIZING

SOCIAL MEDIA DATA
The ability to illustrate and model individual behavior using social media
data shows promise in the design and
deployment of next-generation wellness-facilitating technologies. Privacypreserving software applications and
services can serve as early warning
systems, providing personalized alerts
and information to individuals.
Beyond monitoring behavioral
trends in real time, social media-based
measuressuch as degrees of activity,
emotional expression, etc.can serve
as a personal diary-type narrative
resource logging behavioral fingerprints over extended periods of time.
The application might even assign
health-risk scores to individuals based
on predictions made about forthcoming changes manifested in their behavior. In operation, if the inferred likelihoods of forthcoming changes surpass
a threshold, the individual could be
warned or engaged, and information
might be provided about professional
assistance and/or the value of social
support from friends and family. In
short, we hope analytic approaches
based on social media data can play a
role in helping individuals find timely
and appropriate support from health
care professionals and others.
Concerns regarding individual privacy, including certain ethical considerations, may arise with this form of
analyses of social media, as they ultimately leverage information that may
be considered sensitive given their focus on behavior and health. I envision
the systems described to be designed
as privacy-preserving applications that
are deployed by and for individuals,
thereby honoring the sensitive aspect
of revealing different types of healthrelated information to them.
Closely intertwined with this privacy issue is the challenge of interventions. Can we design effective interventions for people whom we have
inferred to be vulnerable to a certain
illness in a way that is private, yet still
raises awareness of this vulnerability
to themselves and trusted others (doctors, family, friends)? In extreme situations, when an individuals inferred
vulnerability to an illness with risktaking attitudes is alarmingly high
(e.g., self-harm-prone individuals),
what should be our responsibility as
a research community? For instance,
should there be other kinds of special interventions where appropriate
counseling communities or organizations are engaged? In short, finding
the right types of interventions that
can actually make a positive impact
on peoples behavioral state while
abiding by adequate privacy and ethical norms is a research question on
its own. We hope this article triggers
conversations and involvement with
the ethics and clinician community to
investigate opportunities and caution
in this regard.
Beyond interventions, there is need
for work on educating users about the
privacy risks of sharing sensitive information online that can potentially be
linked to their health. Participants social media use suggests they might not
be aware of the implications of some of
these sharing practices, indicating they
may be unaware of how some advertising companies may be collecting and
distributing their information. Even
though a lot of the health inferences
found in prior research are derived
from implicit patterns in activity and
content, the ability to derive any information about a persons health from a
public venue like Twitter may have seX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
rious repercussions (e.g., higher insurance rates, denial of employment, etc.).

Developing interfaces that remind users of these risks (e.g., triggering an
are you sure? dialogue when detecting sensitive information being entered into a post) is an important area
of future exploration by the social computing research community.
A NEW AND EVOLVING FIELD

To summarize, it is important to
bring the potential of social media to
the fore, so as to leverage the benefits
of this new data source in order to enhance the quality of life for people.
Further, this will stimulate discussion and awareness of the potential
role policies could play in supporting the identities and practices that
individuals suffering from certain
illnesses develop in the face of social
disadvantage.
The role of social media and its potential in understanding health behaviors is a relatively new and evolving
phenomenon, one which society is only
beginning to assess and understand.
Because social media are mostly created and controlled by end users, the
opportunity for surveillance and prevention can be extended to all users.
One way to do this could be the public
promotion of direct and easy avenues
for people to access help through social
media sites. Lastly, proactively using
social media to increase public awareness of and education on health issues
is a logical modern public health approach that has the potential to save
many lives.
References
[1]
Social Media Update, 2013. Retrieved from http://

www.pewinternet.org/2013/12/30/social-mediaupdate-2013/.
[2] De Choudhury, M., Gamon, M., Counts, S., and

Horvitz, E. Predicting depression via social media.
In Proceedings of the Seventh International AAAI
Conference on Weblogs and Social Media (Boston).
AAAI, 2013, 128137.
[3] Park, M., Cha, C., and Cha, M. Depressive moods
of users captured in Twitter. In Proceedings of the
SIGKDD Workshop on Healthcare Informatics (HIKDD) (Philadelphia). ACM, New York, 2012, 18.
[4]
Sadilek, A., Kautz, H., and Silenzio, V. Modeling

spread of disease from social interactions.
In Proceedings of the Sixth International AAAI
Conference on Weblogs and Social Media (Dublin,
Ireland). AAAI, 2012, 322329.
[5] Paul, M., J. and Dredze, M. You are what you tweet:
Analyzing twitter for public health. In Proceedings of
the Fifth International AAAI Conference on Weblogs
and Social Media (Barcelona, Spain). AAAI, 2011,
265272.
[6]
Culotta, A. Towards detecting influenza epidemics

by analyzing Twitter messages. In Proceedings
of the First Workshop on Social Media Analytics
(Washington DC). ACM, New York, 2010, 115122.
[7] Moreno, M., Jelenchick, L., Egan, K., Cox, E., et al.
Feeling bad on Facebook: Depression disclosures
by college students on a social networking site.
Depression and Anxiety 28, 6 (2011), 447455.
[8]
Kotikalapudi, R., Chellappan, S., Montgomery, F., Wunsch, D., and Lutzen, K. 2012. Associating depressive
symptoms in college students with internet usage
using real Internet data. IEEE Technology and Society
Magazine 31, 4 (2012), 7380.
[9] De Choudhury, M., Counts, S., and Horvitz, E.

Predicting postpartum changes in emotion and
behavior via social media. In Proceedings of the
SIGCHI Conference on Human Factors in Computing
Systems, (Paris). ACM, New York, 2013, 3267-3276.
[10] Ertin, E., Stohs, N., Kumar, S., Raij, A., alAbsi, M.,
and Shah, S. AutoSense: Unobtrusively wearable
sensor suite for inferring the onset, causality, and
consequences of stress in the field. In Proceedings
of the Ninth ACM Conference on Embedded
Networked Sensor Systems (Seattle). ACM, New York,
2011, 274287.
[11] Shiffman, S., Stone, A. A., and Hufford, M. R.
Ecological momentary assessment. Annual Review
of Clinical Psychology 4 (2008), 132.
[12] Jourard, S. M. Healthy personality and selfdisclosure. Mental Hygiene 43 (1959), 499507.
[13] Vogel, D. L., and Wester, S. R. To seek help or not to
seek help: The risks of self-disclosure. Journal of
Counseling Psychology 50, 3 (2003), 351.
[14] Ellis, D., and Cromby, J. Emotional inhibition: A
discourse analysis of disclosure. Psychology &
Health 27, 5 (2012), 515532.
[15] Pennebaker, J. W., and Chung, C. K. Expressive
writing, emotional upheavals, and health.
Foundations of Health Psychology, 2007, 263284.
[16] Ramirez-Esparza, N., Chung, C. K., Kacewicz, E., and
Pennebaker, J. W. The psychology of word use in
depression forums in English and in Spanish: Testing
two text analytic approaches. In Proceedings of the
Second International AAAI Conference on Weblogs
and Social Media (Seattle). AAAI, 2008, 102108.
[17] Joinson, A. N. Self-disclosure in computer-mediated
communication: The role of self-awareness and
visual anonymity. European Journal of Social
Psychology 31, 2 (2001), 177192.
[18] Smyth, J. M. Written emotional expression: effect
sizes, outcome types, and moderating variables.
Journal of Consulting and Clinical Psychology 66, 1
(1998), 174.
[19] Rodriguez, R. R., and Kelly, A. E. Health effects of
disclosing secrets to imagined accepting versus non
accepting confidants. Journal of Social and Clinical
Psychology 25, 9 (2006), 10231047.
[20] Johnson, G.J. and Ambrose, P.J. Neo-tribes: The
power and potential of online communities in
healthcare. Communications of the ACM 49, 1
(2006), 107113.
[21] Eysenbach, G., Powell, J., Englesakis, M., et al. Health
related virtual communities and electronic support
groups: Systematic review of the effects of online peer
to peer interactions. BMJ 328, 7449 (2004), 1166.
[22] Sillence, E., Briggs, P., Harris, P., et al. How do
patients evaluate and make use of online health
information? Social Science and Medicine 64, 9
(2007), 18531862.
Biography
Munmun De Choudhury is currently an assistant professor
at the School of Interactive Computing at Georgia Tech
and a faculty associate with the Berkman Center for
Internet and Society at Harvard. His research interests are
in computational social science, with a specific focus on
reasoning about our health behaviors and well-being from
social digital footprints.
2014 ACM 1528-4972/14/12 $15.00
27
feature
Here Comes
the #Engagement:
A serious
health initiative
made trendy
Creating a user experience to communicate the seriousness of HIV
prevention and awareness can be both educational while entertaining.
This combination along with a sense of cultural influence helps to
both attract and engage millennials.
By Fay Cobb Payton and KaMar Galloway
DOI: 10.1145/2691362
ay 2006. Anticipating the usual visit from students seeking advice about final
projects, discussing some stubborn bug in their SQL code, or poring over the
cryptic results of data analytics software, a professor sat working in her office The
day took a different turn, however, when a female student entered with a worried
look on her face. Dr. Payton, my boyfriend tested positive for HIV. I do not want to become a
statistic.
In 2011, out of the 49,272 total cases of HIV diagnosed in the United States, an estimated
47 percent were among African Americans. Of the 2,294 cases diagnosed among teenagers,
67 percent were African-American teens. More recent data shows a growing trend of HIV
infections among persons between 13
and 24 years old [1]. These statistics indicate HIV remains a potent threat to
those who are young and college aged,
as well as the Black population. This
warrants increased attention from the
general public and policy makers. That
fateful spring day marked the begin-
28
ning of a journey on the road to MyHealthImpactNetwork.org.

But what does HIV have to do with
creating an online experience that is as
engaging as it is informative? The answer lies in understanding that different socio-ethnic groups perceive and
react to online information differ-
ently. The Internet is expected to play

a vital role in reducing long-standing
health inequalities. Its ability to reach
those suffering from stigmatized
medical conditions and those interested in learning more about such
illnesses is improving every day [2].
However research scholars caution
that in normalizing discourses such

as the digital divide, ethnic minorities, especially African Americans, are
presented as being deficient in computer skills and their ability to utilize
online resources [36]. Such deficit
philosophy models rely on the faulty
assumption that information available online is tailored to appeal to all
psychological and cultural tastes. For
instance, a study of African-American
womens attitude toward the Internet
as a resource of health information
revealed participants differentiated
between Internet access and content
creation, with the latter largely perceived as being dominated by white
culture. Menus, wording, and even
navigation impacts how users view,
form, and experience online identiX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
ties. This often represents how content providers reproduce inequality

and fail to accommodate the cultural
perspectives of ethnic minorities [7].
With these considerations in mind,
our team of research scholars and college students pondered over how to
better tune the content of health information to make it resonate with
college-aged millennials, specifically
Black women. HIV is a serious illness
and carries a strong social stigma.
Quite understandably, the subject is
hardly ever associated with the concept of entertainment. Nevertheless,
fun experiences are more attractive,
offer intrinsic rewards, impact engagement, and have a powerful influence on how people persist in using a
system [8].
SOCIAL EDHEALTH-TAINMENT
We decided no design or content
should be devoid of fun. Rather, it
should be the very vehicle that drives
participation. In our case, fun implied a culture of socio-technical edhealth-tainment [9]. That is, a user
experience (UX) that simultaneously
educates and entertains. Furthermore, we sought to account for the
user communitys social identity and
cultural nuances. Our goal was to create a user experience, not just an IT artifact. In doing so, we hoped to create
a fun working environment. We were
dealing with the serious topic of HIV
awareness and prevention information. But by rethinking creative ways
to disseminate information, not only
were we generating engaging experi29
feature
Figure 1. Twitter use by race and age.
Young African Americans have high levels of Twitter use
% of internet users in each age group who use Twitter
White
All
Internet
Users
16
22
28
18-29
40
21
30-49
50-64
Black
21
10
9
Pew Research Centers Internet Project July 18September 30, 2013 tracking survey.
N=6010 adults ages 18+. For results based on internet users, n=3,617 for whites
and n=532 for African Americans.
Figure 2. The MyHealthImpactNetwork.org landing page using two screen

captures.
ences, but, more importantly, delivering relevant matters.

For millennials, social means fun;
engaging with friends and family
over social networks has become an
essential part of the daily recreation
of the younger generation. The widespread use of social media has made
it convenient to understand the living,
working, and playing habits of this
demographic. College-aged AfricanAmerican millennials have a strong
digital presence and are particularly
heavy users of Twitter. Forty percent of
African-American Internet users aged
1829 say they use Twitter; a figure that
is 12 percentage points higher than the
comparable figure for white people,
only 28 percent of whom are Twitter
users [10, 11].
Therefore, given the projects focus
on young African Americans, Twitter was the best tool for disseminating
health messages. This notion was confirmed by the qualitative data collected
[12]. The studys findings from 40 Black
women indicated lack of trust, stigma
ascribed to HIV, and misconceptions
create communication barriers. Therefore the empowering ethos of nothing
about us, without us, engendered by
the culturally aware design of the UX,
resonated particularly well with Black
women [12]. The team ultimately adopted for students, by students as the slogan for our project to reflect this spirit
of empowerment.
Several
additional
findings
emerged from the qualitative data.
Even among stigmatized health conditions, web-based user experiences
can interject fun while communicating serious messages to grab the
target audiences attention. We were
able to establish reciprocal fun for the
user by creating cultural relevant messages via social media [13]. For our designers, fun manifested itself in team
interactions and meeting our intended audience in the social and physical
spaces where they reside. As researchers it was important that we understood the socio-technical impacts of
technology in all populations, and in
particular, those under-represented
and underserved.
BUILDING A COMMUNITY
There was eager excitement as our
30
team worked on the final stages of the

project. Deciding the color theme for
the landing page of MyHealthImpactNetwork.org was a particularly hot
topic of debate. We wanted to catch
the users attention at first glance and
increase the comprehensibility of our
content. We also wanted to take advantage of the fact that people have common associations with colors, positive
and negative.
The initial design envisioned a
neutral color theme that would appeal to both men and women, that
latter group being our primary focus. The first set of landing page was
predominantly red. But the feedback
from the design team, which was
equally divided by gender, was mixed
at best: Do we really want red which
is the HIV/AIDS ribbon color? Another team member added, Some say it
means that you have tested positive;
that is not the message we want to
send or brand we are taking on? We
are focused on prevention and awareness. The next choice was blue, but
the debate continued. Although blue
is a color that appeals to men and
women, we couldnt agree: We dont
want to lock females out because the
world says blue is for guys. The team
eventually compromised by choosing purple, a color halfway between
red and blue. It works for females.
I can see this, said one male team
member. But would it work for men?
To which another male team member
explained: We can get with that [the
purple]. It gets our attention, and we
remember that the initial project focus is NOT about us [men]. It is not
about us, but the information will
help us as well.
Once the design was finalized, our
next concern was content. A couple
of young team members, male and
female, were especially excited about
the opportunity to see and, in some
cases meet, celebrity speakers, musicians, and entrepreneurs at a local
health event, which they covered for
MyHealthImpactNetwork.org.
The
students captured digital content
and gathered HIV prevention and
health information, which they later
blogged about. During the post-event
review, we discussed how to best use
the content and event experience. As
we brainstormed, the male attendee

stated: I enjoyed myself. Black women were everywhere. It was good to
see the positivity, but I admit I was a
little intimidated. There were not a lot
of brothers around, but the few I saw,
we all seemed to feel the same way. We
gave each other the head nod. The female attendee replied: I never would
have thought Ken [name changed]
would be intimidated by women. Was
it too many in one place for you? He
replied, I got comfortable after I understood the situation. This was an
important exchange in that it demonstrated how empowerment is central
to design and how offline interactions
can inform the online health messaging.
inclusive of ethnic and gender voices,

too often void or minimal in the technology space, to shape to both create,
consumer and disseminate the health
messages. This is the essence of social
edhealth-tainment.
The National Science Foundation
grant IIS-1144327 supports this research.
CONCLUSION
In addition to Twitter, a blog and a
YouTube channel were created when
the site was launched. The blog gives
individual team members and guest
writers an opportunity to voice their
opinions on critical topics such as the
stigma associated with HIV; provide
a male perspective on health issues;
discuss health-related messages in
hip-hop music, popular culture, and
the news; and prompt readers to take
action for social justice. In addition,
users are able to interact with the
research team, giving myHealthImpactNetwork.org a personal feel. The
YouTube channel enables viewers to
see the team in action. Music and art,
which are hidden talents for some
team members, have been incorporated in our social media channels
and videos, helping further our social
edhealth-tainment approach in circulating health messages.
HIV is a hard topic to discuss. MyHealthImpactNetwork.org has shown
that information about serious health
conditions can be made more interesting to a young audience by interspersing it with the right degree of levity,
communicating via social media, and
using a carefully designed user experience. The coolness of the content results from the minimal use of
medical jargon, plenty of hooks to pop
culture and news events, quick and
effective communication, and a user
experience created for students, by
students. This user experience is also
[5] Kvasny, L. and Payton, F. C. African Americans and

the igital divide. In Encyclopedia of Information
Science and Technology, 2nd Edition. M. KhosrowPour (Ed.). Idea Group Publishing, Hershey, PA,
2008, 7882.
References
[1]
Center for Disease Control and Prevention (2013).

Statistics on HIV Surveillance. http://www.cdc.gov/
hiv/pdf/statistics_surveillance_adolescents.pdf
[2] Berger, M., Wagner, T.H. and Baker, L.C. Internet use
and stigmatized illness. Social Science & Medicine
61, 8 (2005), 18211827.
[3] Brock, A. A belief in humanity is a belief in colored
men: Using culture to span the digital divide.
Journal of Computer Mediated Communication 11,
1 (2007), 357374.
[4]
[6]
Kvasny, L. The role of the habitus in shaping

discourses about the digital divide. Journal of
Computer Mediated Communication 10, 2 (2005).
Selwyn, N. Reconsidering political and popular

understandings of the digital divide.
[7] Nakumara, L. Cybertypes: Race, Ethnicity, and

Identity on the Internet. Routledge, New York, 2002.
[8]
Carroll, J.M. and Thomas, J.C. Fun. SIGCHI Bulletin 1,

3 (1988), 2124. New Media and Society 6, 3 (2004),
341362.
[9] Siek, K.A. What are our responsibilities when

designing sociotechnical health interventions?
Interactions 18, 5 (2011), 2023.
[10] Nielsen. African-American consumers: Still vital,
still growing 2012 Report. The Nielsen Company.
2012.
[11] Pew Internet Research (2014), African American
and Technology Use.
[12] Payton, F.C., Kvasny, L. and Kiwanuka-Tondo, J.
Seeking and perceiving online HIV prevention
information: Black female college students
perspectives. Internet Research 24, 4 (2014),
520-543.
[13] Payton, F.C. and Kiwanuka-Tondo, J. Contemplating
public policy in AIDS/HIV online content, then
where is the technology spirit? European Journal
of Information Systems 18, (2009), 192-204.
Biographies
Fay Cobb Payton directs MyHealthImpactNetwork.org, a
social network experience that focuses on health disparities
and social media technology interventions. She is the
author of Leveraging Intersectionality: Seeing and Not
Seeing, an anthology of her research on STEM education and
experiences in both academe and corporate environments.
Dr. Payton is an editor for Health Systems, an OR Society
journal, and is an associate professor of information
systems at North Carolina State University. She received
the 2013 National Coalition of Women in Information
Technology (NCWIT) Undergraduate Mentoring Award. She is
the corresponding author and principal investigator for this
article. Dr. Payton is an active Twitter user; @drfayonline.
KaMar Galloway is a graduate of the Computer Science
Department at North Carolina State University and an active
sneaker collector. Combining his love for technology and
footwear, he is interested in the birth of wearable computing
and the impact it will have on making healthy decisions.
He is interested in online platforms that are informative,
welcoming and can assist the African-American community.
2014 ACM 1528-4972/14/12 $15.00
31
feature
Challenges in
Personal Health
Tracking: The data
isnt enough
Increasingly, personal health data can be tracked and integrated from
numerous streams quickly and easily, but our feedback lingers in the
land of show the user a graph and hope. How can we help people
make sense of personal health data?
By Matthew Kay
DOI: 10.1145/2678024
bathroom scale may not have the same

API or data format as your pedometer
projects like S Health (http://content.
samsung.com/us/contents/aboutn/
sHealthIntro.do) and Apple HealthKit
(https://developer.apple.com/healthkit/)
are breaking down these barriers by
encouraging developers to support
these APIs or be left behind.
What we are left with is the question of what to do with all that data.
It remains the case that most health
apps take some variant of the show
32
em a graph and hope approach: Surely people will notice the correlations
between their different data streams
and draw appropriate conclusions if
we just show them the data! Nevermind that one third of Americans have
low graphical literacy [1] and would be
unlikely to make valid inferences from
supposedly straightforward graphical
data. Asking people to make statistical
judgments from graphical datasuch
as identifying correlationswithout
providing scaffolding too often leads
to bewilderment, or worse, the drawing of spurious conclusions. People

even trained statisticiansare particularly good at seeing patterns in data
where there are none.
Some of the core challenges for
personal health informatics, then, lie
in finding and presenting health data
to people in useful and meaningful
waysit isnt enough to bring the data
into one place. This leads us to ask
questions like: What data do we actually need? How can we summarize it
Image by Syda Productions
ersonal informaticsand particularly personal health trackingis booming.

Apple and Samsung have both announced apps for aggregating personal health
data, such as step counts or sleep quality, into one unified interface. This is the
first step in achieving a long-standing vision in personal informatics. Giving people
one place to see visualizations of all of their health data. A lot of the low-level technical
challenges for this vision are being solved: We can sense myriad aspects of healthfrom
step count, to sleep quality, to stress, to respiratory function, to heart rate variability.
As users and developers, we can also access this data more easily than before. While it is
sometimes cumbersome to get all of that data out of each providers siloyour Wi-Fi
33
feature
Figure 1. Lullaby deployed on a bedside table. Visible here are a touchscreen tablet
mounted in a stand for easy access from the bed and the sensor box with pivoting
sensor enclosures. The sensor suite itself is about the size of a bedside lamp.
in a way that aids understanding? I

explored some of these questions in
the context of sleep in a project called
Lullaby, a research prototype that includes a suite of environmental sensorssound, light, temperature, and
motionto help people assess the
quality of their sleep environments
(see Figure 1) [2]. In that project, we
built a system to track how peoples
bedroom conditions (such as light levels or air quality) impact their sleep
quality, and explored different methods of presenting that data in meaningful ways to people.
IS YOUR BEDROOM MAKING

YOUR SLEEP WORSE?
Research has shown environmental
factors can be a major cause of poor
sleep quality and interrupted sleep [3];
for example, a room that is too warm,
has improper lighting, is noisy, or has
poor air quality can disrupt sleep.
While some of these environmental
factors are observable, others may be
subtle or difficult to recognize. Thus,
individuals who have poor sleep quality
often have trouble evaluating the cause
or severity of their sleep difficulties.
Figure 2. A screenshot from the Lullaby feedback app showing a single night of sleep data.
34
While clinical sleep centers can

evaluate an individuals sleep quality
effectively, these evaluations do not
occur in individuals actual homes.
Thus, a sleep lab cannot directly
identify environmental factors in
the home that contribute to reduced
sleep quality. At the same time, commercial personal informatics devices
like the Fitbit can help identify when
a person has had poor sleep or when
they awaken at home, but generally
dont measure environmental conditionsthey report measures like the
proportion of time in bed actually
spent asleep (sleep efficiency), and
when during the night sleep has been
disturbed. These measures provide
some indication of sleep quality, but
cannot give concrete guidance for
sleep environment improvement.
Ubiquitous computing technology
that helps people track both their sleep
habits and environmental factors that
affect sleep quality could help people
identify why their sleep was interrupted, not just when. That was the goal of
Lullaby. Using a tablet device kept by
the users bed, Lullaby displays this
environmental data together with data
from an off-the-shelf sleep-tracking
device, like a Fitbit. It aims to help
people better understand their sleep,
to understand what goes on in their
sleep environment while they are unconscious, and to help them make improvements in their sleep habits.
Lullaby was motivated by discussions with sleep doctors: Some patients report sleeping poorly at home,
but sleep just fine when they come in
to the sleep lab for an assessment. Of
course, if there are any problems being
caused by factors in their bedroom at
home, these issues will not show up
in a sleep lab. This is the perfect opportunity for personal health tracking to help answer questions that were
difficult or impossible to answer in
the past. To help people find answers
to these questionsIs my bedroom
temperature affecting my sleep?we
need to understand how to put together feedback from a device like Lullaby.
CAN PEOPLE MAKE SENSE

OF SLEEP DATA?
We conducted a feasibility study of Lullaby to better understand how people
Designers should
think about how
to design better
feedback devices
given their inherent
uncertainty and the
properties of the
data being collected.
might use Lullaby and how they interpret the data they see on the device (see
Figure 2). There are particular challenges associated with building applications like this when the data is recorded while people are unconscious.
The domain of sleep is one where
events of interest are not known by users until well after their occurrence
until the time at which the user goes
looking for such events. As a result,
users must sift through data with little
or no knowledge of what they seek or
when it occurred, so helping them discover salient data is very important.
To aid this discovery, we gave users
a wider context in which to view their
data by highlighting data that is out of
recommended ranges from the sleep
literature (e.g., too hot or cold, too
noisy, etc.). We also showed all collected data together, chunked by sleep period, and allowed people to play back
sound and infrared images of their
sleep alongside the data to provide a
more concrete frame of reference.
People found this unconscious
data compelling: Imagine watching
yourself sleepor for one participant,
sleepwalk. Another person found she
coughed regularly in her sleep by observing consistent spikes in audio
data. This illustrates the potential of a
system like Lullaby. Chronic coughing
is a symptom of sleep apnea, a condition that is widely undiagnosed largely
because its immediate symptoms are
difficult for the sufferer to observe (one
of the very reasons that motivated us to
develop Lullaby in the first place).
Compelling aspects of the data
aside, there are significant challenges
to helping people draw meaningful

conclusions here. One of the goals of
Lullaby was to help people identify
the things that are causing disruption
to their sleep. Some disruptors might
be easy to see just by looking at the
graphs in tandem with the audio-visual stream, such as a co-occurrence between awakenings and motion caused
by cats entering or exiting the bed.
However, more subtle causes may be
difficult to determine just by reviewing
the data manually. For example, users
may want to know if they awaken more
frequently while the temperature is
warmer over a period of several nights.
With enough captured data, Lullaby
could help identify such relationships
by running statistical analyses on the
data in order to produce higher-level
summaries, but expecting people
to find such correlations by looking
through time-series graphs themselves is probably unrealistic. We need
to consider another approach.
THE POWER OF NATURAL

LANGUAGE SUMMARIES
To investigate the potential for higherlevel summaries and inferences, we
followed up with some participants
after the study ended to present mockups of possible future feedback interfaces. These mock-ups ranged from
scatterplots of potentially related factors, to aggregate statistics of various
measures (for example, the amount
of time a given sensor spent out of its
recommended range from the sleep
literature), to one-sentence summaries
of factors influencing sleep (for example, Over the past two weeks, higher
temperature has been associated with
worse sleep.).
Participants responded particularly
well to the single-sentence summaries. In one participants case, Lullaby
had found a possible correlation between higher temperature in the bedroom and sleep disturbance (when the
temperature went up, his sleep quality tended to do down). Presented as a
scatterplot with a trend line, he found
this to be an interesting finding, but
given a single-sentence natural language summary, he became excited:
That would be really cool.
More generally, natural language
has many potential advantages in
35
feature
CACM_TACCESS_one-third_page_vertical:Layout 1
ACM
Transactions on
Accessible
Computing
This quarterly publication is a

quarterly journal that publishes
refereed articles addressing issues
of computing as it impacts the
lives of people with disabilities.
The journal will be of particular
interest to SIGACCESS members
and delegates to its aliated
conference (i.e., ASSETS), as well
as other international accessibility
conferences.

www.acm.org/taccess
www.acm.org/subscribe
36
6/9/09
1:04 PM
Page 1
sense-making. It is intuitive and easily

understood, and more explicitly identifies connections in data for people.
While a statistician might find a scatterplot easy to interpret, most people
need more scaffolding than this. Spotting correlations in time-series data
is similarly difficult, and the almost
purely graphical focus of many existing personal informatics applications
is insufficient in the face of the realities of statistical and graphical literacy
levels in the general population.
The results from the Lullaby project
made me realize the need for a better
understanding of personal health data
literacy. I decided to simplify the problem. Lullaby has many interacting data
streams, making it difficult to tease
apart the many facets of health data
literacy. I turned to another domain,
perhaps the simplest, most ubiquitous
example of a personal health sensor:
the bathroom weight scale.
A TEST CASE FOR DATA LITERACY

The bathroom scale is arguably the
most ubiquitous health sensor of all,
and it is effective, too: Studies have
shown that frequent weigh-ins help
maintain weight loss [4]. At the same
time, many people have a love/hate relationship with their scale, and dread
standing on it. Despite its centrality to global health and wellness, the
bathroom scale interface has barely
changed since it was first introduced
about 100 years ago: It still produces a
single value representing ones weight
at the moment of measurement. Digital displays have replaced the analog
needle, coarse measurements of body
fat have been added, and some scales
log data for offline review. However,
the singular data point is still the main
display and is often the only information presented at the time of weigh-in.
Most scales answer just one question
What do I weigh right now?which
may not be the best framing for weight
data. However, its ubiquity makes it a
good test case for literacy: If we cant
get the scale right, what hope do we
have for more complex health data
streams? We conducted a series of
studies [5] to investigate how the scale
interface impacts data understanding.
We found several issues with current scales that work against an effec-
tive understanding of weight data. For

example, digital scale readouts convey an unrealistic level of precision,
negatively affecting user perception.
Scales often present data at an apparent resolution of a tenth of a pound or
less, when fluctuations of a few pounds
within a single day are typical of normal individuals. This level of apparent
precision can cause people to focus on
small, insignificant changes in their
weight. Experts in weight change (such
as dieticians or personal trainers) will
try to mitigate the effect of these fluctuations, both by telling people to always
weigh under similar conditions and
by educating people about why weight
can fluctuate over the short term. In
interviews with such experts, one described clients as getting the horrors
when they feel like their weight moves
in an undesirable direction even a
small amount. People will fixate on
and form an identity around a particular weighta particular numberthey
want to be. This is only reinforced by
an interface that communicates little
else but a number.
We conducted an online survey of
more than 800 scale users and quizzed
them on their knowledge of weight
fluctuation. We found respondents
with a less sophisticated understanding of how weight fluctuates during the
day were less likely to trust their scales.
This is exacerbated by the fact that the
scale interface makes no attempt to
inform users about how weight fluctuates: The single data point reflects
nothing about weight change over time
or the accuracy and precision inherent
We can use more

appropriate
sensors, with more
intelligent feedback,
to give better,
clearer answers to
underlying health
and wellness
concerns.
to that measurement. It is no wonder

people fixate on single numbersit is
the only thing they are given.
In a study of consumer reviews of
scales, we found many people would
rate scales poorly based on some assessment of their accuracy, but often
the expectations of accuracy people
had were unrealistic. For example,
some people had expectations of inter-scale reliability that are not even
realistic in hospital settings: One person complained their scale was off
by about a pound from their doctors
scale. However, even scales in hospitals can reasonably be expected to
differ by as much as 1.5 pounds when
measuring the same person [6].
What we are left with is a device that
gives data to users without setting reasonable expectations for the accuracy
of that data, without providing context
to the data (such as trends over time),
and without attempting to provide the
scaffolding of education about weight
fluctuation over time. It is no wonder
that people often react negatively to
the data they see, or have an aversion
to stepping on the scale. Further, as
scales are part of a larger class of increasingly ubiquitous health feedback
devices, which provide single-point, instantaneous measurementssuch as
body fat estimators, thermometers, pedometers, and blood pressure cuffs
that we have not gotten the scale interface right bodes poorly for the others.
SO WHAT SHOULD THESE

INTERFACES LOOK LIKE?
Between Lullaby and the bathroom
scale we see similar issues encountered across the space of personal
health informatics: Making sense of
data is hard, and proper support for it
is not there yet in many existing systems. At the same time, we have made
good progress on technical issues.
Sensing and data infrastructure have
both improved dramatically, leading
us to turn back to questions of what to
sense and what to do with the data now
that we can get it.
I believe, for example, a better scale
can be designed without investing in
more expensive equipment, better calibration, or even clearer instructions
for obtaining better data (e.g. to always
use the scale on a hard surface), by inX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
Users must sift

through data
with little or no
knowledge of what
they seek or when it
occurred, so helping
them discover
salient data is very
important.
stead handling the resulting data and

consequent user feedback in a more
considered way. Scales are already
ubiquitous, cheap, and fairly accurate;
greater gains may be had by pushing
the state of the art in feedback. Designers should think about how to design
better feedback devices given their inherent uncertainty and the properties
of the data being collected.
There are many things we can do
to address these issues. For example,
device feedback should avoid false
precision. Instead, give people a sense
of how precise the data is and how
much fluctuation they can interpret
as meaningful change versus noise. A
scale that adopts a more sophisticated
model of the underlying data could
also set user expectations more explicitly. For example, a scale might report estimates using natural language
explanations: We estimate your daily
weight to within 3 pounds since your
weight typically fluctuates about that
much during the day, or may involve
graphical depictions of weight variability. Combining multiple data
streams, health devices could go even
further: When you sleep poorly your
weight goes up. Frank Bentley has
done excellent work exploring this approach [7].
More generally, I hope future designs of ubiquitous health sensors
thermometers, blood pressure cuffs,
blood glucose monitors, etc.adopt
a more sophisticated approach to
feedback. I expect similar issues
such as gaps in knowledge of the ac-
curacy of a device or what constitutes

typical fluctuationwill manifest in
these domains. Better reflecting the
underlying model and educating users about uncertainty will likely be
crucial here as well. Both graphical
and natural language presentations
of data must be sensitive to peoples
mental models of the data being communicated. As we continue to push
low-cost novel health sensing into the
wild, it is important to consider how
the accuracy of these systems affects
users, and how much we can gain
from more intelligent feedback.
Finally, we have an opportunity
to revisit the underlying health questions users want answered, what data
we can use to answer them, and how to
help interpret that datarather than
falling back to the nearest convenient
measure (e.g., weight, sleep quality,
step count) and simply reporting individual data points without context
or interpretation. Instead, we can use
more appropriate sensorswith more
intelligent feedbackto give better,
clearer answers to a persons underlying health and wellness concerns.
References
[1]
Galesic, M. and Garcia-Retamero, R. Graph literacy: A

cross-cultural comparison. Medical Decision Making:
An International Journal of the Society for Medical
Decision Making 31, 3 (2011), 44457.
[2] Kay, M., Choe, E.K., Shepherd, J., et al. Lullaby: A

capture and access system for understanding the
sleep environment. Ubicomp 12, (2012).
[3]
Kryger, M.H., Roth, T., and Dement, W.C. Principles

and practice of sleep medicine. W. B. Saunders Co.,
Philadelphia, 2000.
[4]
VanWormer, J.J., Linde, J.A., Harnack, L.J., Stovitz,

S.D., and Jeffery, R.W. Self-weighing frequency is
associated with weight gain prevention over two
years among working adults. Int J Behav Med 19, 3
(2012), 351358.
[5] Kay, M., Morris, D., Schraefel, M., and Kientz,

J.A. Theres no such thing as gaining a pound:
Reconsidering the bathroom scale user interface.
Ubicomp 13, (2013), 401410.
[6]
Goldberg, R. and Hebbard, G. How accurate are

hospital scales? The Medical Journal of Australia
194, 12 (2011), 665.
[7] Bentley, F., Tollmar, K., Stephenson, P., et al. Health

mashups: Presenting statistical patterns between
wellbeing data and context in natural language to
promote behavior change. ACM Transactions on
Computer-Human Interaction 20, 5 (2013), 127.
Biography
Matthew Kay is a Ph.D. student in computer science and
engineering at the University of Washington studying
visualization and feedback in ubiquitous computing
systems. He is particularly interested in improving the
design of user interfaces backed by data with inherent
uncertainty, such as those in personal informatics.
2014 ACM 1528-4972/14/12 $15.00
37
feature
Did I Take
My Meds
Today?
People tend to believe they are more aware of their own health behaviors
than they really are. In this article, we present technologies that employ
ubiquitous home sensing to support awareness of healthy habits.
By Matthew L. Lee
DOI: 10.1145/2676574
normally part of your routine. Similarly, actions people take to manage

their health, such as taking medications or eating a healthy diet, are
often performed as part of a regular
routine with actions that can easily
be overlooked.
To address a health behavior problem like missed medication or a series
of unhealthy food choices, people must
first be made aware of the problem
38
and the context in which it occurred

to prevent it from happening again.
As individuals are encouraged to take
the drivers seat in their own health
journeys, they need tools that not only
reinforce regular healthy behaviors,
but also make them aware of when and
how their behaviors can be improved.
The once fairy-tale vision of ubiquitous monitoring of our everyday actions has now reached the common
consumer in the form of commodity apps (e.g., Lift, Samsung sHealth,

FitBit) easily downloaded to mobile
smart devices. They also come in the
form of affordable gadgets found at any
shopping mall, which can be installed
in the home simply by plugging them
in (e.g., Nest, GlowCap, CubeSensors).
Monitoring apps and devices demand
little of our attention and quietly accumulate a log of our actions over time.
Image by Tobias Arhelger
eople are creatures of habit. In our everyday lives, we naturally fall into routines
where we, for better or worse, mindlessly perform regular actions that make up
our day. For example, consider your daily drive to work. When you reach your
destination, you follow the routine to turn off the car, exit the car, close the door,
and lock the car doors. However, sometimes the action of locking the door is so automatic
that you might not explicitly remember whether you actually locked the doors. You might
actually have forgotten to lock the doors, but believe the doors are locked because it is
39
feature
Figure 1. Sensor-augmented pillbox unobtrusively monitored medication taking
over 10 months.
closure switches
custom PCB
accelerometer
They allow us to reflect on our actions

to see if they are consistent with our
health goals. These systems capture
a good deal of our actions in context,
but user feedback often consists of a
simple summary of the data shown to
users when they choose to view it. Both
the content and the timing of the feedback is important for supporting an
awareness of (un)healthy habits and
empowering individuals to improve
their habits.
Medication taking is a health action
almost universally considered important by both those who prescribe medications and those who take medications. Taking medications is difficult
because, unlike other daily habits like
bathing or brushing teeth, it is a habit learned later in life. Further, poor
medication adherence is not inherently self-reinforcing because missing
medications often does not have immediately perceivable bodily effects to
the individual. Approximately 2050
percent of patients do not take at least
80 percent of their medications [1]. C.
Everett Koop, former Surgeon General of the U.S. once said, Drugs dont
work in people who dont take them.
Indeed, the primary concern of primary care physicians is whether their
patients take their medications, which
can lead to improved health outcomes
and reduced costs of care [2].
In our research, we aimed to determine whether feedback about how
people take their medications would
help them take their medications more
consistently and correctly [3]. We designed a home sensing system, called
40
wireless radio
microprocessor
dwellSense, which logs peoples medication taking using a sensor-enhanced

pillbox [see Figure 1] while also tracking a number of other tasks important
for independence, including dialing
the phone and preparing meals/coffee.
Unlike medication prompting devices
such as GlowCaps, dwellSense did
not prompt or remind the user to take
his/her medications at a certain time.
Rather, it was designed to 1. support the
users autonomy by allowing the user to
follow his/her own routine to self-initiate medication taking and 2. provide
feedback soon afterwards so that user
can verify the behavior. Timely feedback is an important part of self-regulation [4], the process by which people
self-monitor their behaviors, judge the
quality of the behaviors based on their
own goals, and make adjustments to
improve their behaviors.
dwellSense is a suite of sensors embedded in the home that monitors and
logs instrumental activities of daily
living, including medication taking,
telephone use (dialing and misdialing patterns), and making a pot of coffee (as a proxy for meal preparation).
These specific tasks were selected because they are performed frequently
and require sequenced procedures
indicative of an individuals cognitive
and functional ability. For medication
taking, a pillbox was augmented with
contact sensors to record when each
pillbox door was opened and closed.
An accelerometer in the pillbox also
logged when individuals handled the
pillbox and inverted it to pour out their
pills (a common action for some indi-
viduals). For telephone use, a sensor

placed on the telephone line logged
the sequence and timing of digits dialed, and software looked for patterns
of misdialed calls (interposed digits,
missing/added digits, etc.). A coffeemaker was augmented with various
contact sensors to detect the sequence
of adding water and coffee to the machine, placing the carafe in the machine, and turning it on.
Each sensor device wirelessly sent
its data in real time to a laptop placed
in the home using a Zigbee wireless
network. The laptop uploaded the data
to a remote server, where the data was
processed and visualized on a tabletbased ambient display [see Figure 2].
Using a user-centered design process,
the tablet display showed near realtime feedback on how individuals carried out their everyday tasks based on
the information needs of individuals.
For medication taking, previous field
work found individuals wanted to use
their own routines to take their meds
but sometimes did not recall whether
they took their pills earlier in the day
[5]. To provide feedback, while also
supporting individual routines, the
tablet visualization merely reflected
whether the individual had not yet
taken their meds or had completed
taking their meds, rather than providing proactive reminders or alerts
to prompt medication taking. Users
defined the latest time of day by which
they planned to have taken their meds.
The feedback display showed whether
meds were taken on time (completed before the user-defined time), late
(completed after the user-defined
time), or missing (not yet completed after the user-defined time). The
display also showed the time of pill
taking, and which pillbox doors were
opened, so users could check whether
they had performed their medication
taking promptly and correctly. The
tablet visualization updated at least
once every 30 minutes, providing near
real-time, glanceable feedback to users about their behavior for the current day before resetting every night
while the individual was sleeping.
The dwellSense system was deployed for 10 months to monitor the
medication taking of 12 older adults
who were living alone and taking
multiple medications for multiple

chronic conditions. Each participant
believed taking medications was important; they also claimed they were
taking their meds consistently without any mistakes.
In the first two months of the deployment, no feedback was given to
the participants about their monitored
behaviors. This baseline period provided time for participants to grow
accustomed to being monitored and
behave as they normally would. Beginning in the third month, half the
participants (the feedback group)
received the real-time feedback tablet
display, and the other half (the control group) received no feedback. The
feedback group placed the display in
their living or dining room so that the
display was easily viewed throughout
the day. At the end of the fourth month,
participants in the control group were
shown a graph of their medication
taking data from the preceding six to
eight weeks to ensure both groups understood the type of data that was being monitored. After the eighth month,
the tablet display was removed from
the homes of the feedback group. In
the remaining two months, the dwellSense system continued to monitor
the participants medication-taking
behaviors. The lengthy 10-month study
was intentional; it was designed to give
time for participants to become accustomed to being monitored by sensors and to integrate the feedback into
their everyday routines. Shorter-term
studies of behavior change technology
often show results that point to an immediate novelty effect or performance
effect, leaving open the question of the
longer-term results of the intervention. By the end of the study, more than
5,785 episodes of medication taking
were observed across all participants.
Moreover, the length of the study also
allowed for the well-regarded A-B-A experimental paradigm for the feedback
group, in which the effects of introducing an intervention (the B phase) and
removing the intervention (a return to
the A phase) can be studied.
Medication taking is a self-regulatory process, which means individuals
require timely feedback about whether
they are taking medications according to their personal standards. In
other words, we hypothesized people

need to know when they are making
mistakes so they know how to fix the
problem. The feedback must also be
given promptly so the reason for the
sub-optimal performance can still be
remembered or inferred. The results
of the 10-month study corroborate this
hypothesis. Participants in the feedback group significantly improved in
two ways after the feedback display
was introduced: 1. promptness (taking
medications before their pre-defined
late time, going from 75.1 to 90.8 percent, F[1,67]=11.40 , p=0.001); and 2.
correctness (opening the pillbox door
that matches the current day of the
week, going from 94.2 to 99.2 percent,
F[1,67]=4.18, p=0.0448), [see Figure 3].
In the same period, the control group,
not surprisingly, remained unchanged.
Furthermore, the feedback group also
significantly decreased in the variance
in the time of day they took their meds
(15.5 to 5.0 hours), after the display was
introduced while the control group actually increased, which likely indicates
individuals in the control group were
less conscientious about taking their
medications despite being monitored.
For the simple metric of the adherence
rate (how often medications were taken at all), participants in the feedback
group increased from 94.5 to 98.4 percent, while the control group remained
relatively unchanged (from 93.7 to 92.1
percent), which was not statistically
significant
(F[1,67]=2.33,p=0.131).
Participants in the study actually began the study with a high baseline adherence rate averaging greater than
90 percent, as measured by dwellSense. We observed a ceiling effect
in the feedback group, increasing
close to the 100 percent adherence
rate. In fact, all six participants in the
feedback group had at least one twomonth streak of 100 percent adherence
with the help of the feedback display, a
trend not observed before the feedback
display was introduced.
Participants reported the display
helped them self-regulate their medication taking by increasing their
awareness, identifying errors, and
confirming their memory. Participant
P05 said the display increased her
awareness because, It always tells
you what time you get your medication
then in the meanwhile sometimes it
tells you when you have missed your
medication. That way, its good. P10
remarked the display helped her identify when she made mistakes: Before,
if I made a mistake and forgot it, I
might not notice until the end of the
Figure 2. A tablet-based display shows feedback about how well individuals carry
out their medication taking, phone use, and coffee making. It shows information
about what time meds were taken; whether it was on time, late, or missing;
what pillbox doors were opened; and a glanceable visual rating (right) for the
promptness and correctness of medication taking.
41
feature
Figure 3. The feedback group increased in the promptness (taking meds before the
user-defined late time) of their medication taking after the feedback display was
introduced into their homes, while the control group did not change significantly.
introduced
feedback display
control group
feedback group
Average Promptness
100%
90%
80%
70%
60%
50%
Month
week or something. [With the display]

you notice it, you see it right away.
P01 mentioned the display gave her a
way to check whether she had taken
her evening pills yet: I look up and
I see Oh! I didnt take my evening
pills, so its a nice reminder, its not a
nasty kind of thing, you know. Even
though participants reported the
display helped them be more aware
and identify errors, it did not make
participants realize when they were
performing their medication taking
sub-optimally. In contrast, when the
control group reviewed a graph of the
68 weeks of their medication-taking
behaviors, most found it shocking that
they missed so many medications and
vowed to make a change. However, left
without an easy means of feedback for
self-regulation, the members of the
control group were only able to maintain a temporary and slight increase
in medication taking performance for
a few weeks before reverting back to
their baseline performance.
Individuals need not only an awareness of errors but also confirmation
that they perform behaviors correctly.
Feedback group participant P02 reported he regularly used the display
to verify he took his pills correctly: I
find myself [checking the display] ev42
ery day, to make sure I was on time or

make sure I was right or make sure I
did the thing I was supposed to. P06
also developed this same habit. In fact,
after the display was removed from
their homes, both P02 and P06 reported they had grown so accustomed to
glancing at the display that they still
turned their head to look in the direction of the tablet after taking their pills
or during TV commercial breaks.
The improvements in medication
taking for the feedback group did not
persist after the display was removed.
Measures of adherence, promptness,
and correctness decreased to a level
that was slightly higher (but not statistically significantly higher) than their
baseline levels before they received the
feedback display. Variance in the time
of day also increased to a level slightly
lower (but not significantly lower) than
baseline. Participants relied on the
feedback to assist them in self-regulating their medication taking. Once
the feedback display was removed,
they did not have an easy way to verify
whether they had taken their medications already and to identify when they
made a mistake. Ongoing frequent
feedback seems to be instrumental
for maintaining the improvements in
medication taking behavior.
The results of this study demonstrate providing feedback can be one

effective means for supporting behavior change. This study also suggests
simply providing behavioral feedback
is not enough, rather, the timing and
frequency of the feedback is important
for triggering and sustaining behavior
change. Confronting individuals with
a long-term record of their medication
taking behaviors convinced individuals in the control group they needed
to improve. But without subsequent
reinforcement and feedback about
their actual performance they could
not troubleshoot their routines and
were only able to sustain improved
medication taking for a few weeks. In
contrast, although providing real-time
feedback had only a gradual, subtle
effect individuals awareness of their
medication taking performance, it offered the timely feedback necessary
for self-regulating their daily medication taking. The frequent feedback
allowed individuals to identify when
they have forgotten or made a mistake, which gave them an opportunity
to either correct the problem or avoid
making the same mistake in the next
pill-taking episode. Real-time feedback and longer-term feedback appear
to have different effects on awareness and follow-through. Combining
these types of feedback likely will have
complementary effects that support
behavior change through increasing
awareness and supporting the actual
change in behavior, making this combination an interesting topic of future investigation that can influence
the manifold health apps that rely on
tracking and reflection.
Another important lesson from
this study is technological solutions
need to respect individuals autonomy.
Most individuals in the beginning of
the study reported they performed
their medication taking without any
problems and did not feel the need
for technology to support this seemingly simple task. Instead of introducing the sensors and tablet display as
an assistive technology that reminded
(or even worse, nagged) individuals to
take their medication according to a
rigid schedule, the technology was introduced as a safety net that could be
ignored or used in whatever way the
individual deemed useful to them. The

display required neither any button
presses nor any other explicit interactions; instead, it simply offered an
easy-to-read, up-to-date information
display, driven by the natural interactions individuals already did (e.g., open
and close the pillbox). Users could consult it to verify their medication taking
with the system and use their own selfregulatory processes to adjust their
daily routines to take their medications more promptly and correctly.
Older adults, just like younger
adults, often do not believe they require help to do simple things even
though their physical and cognitive
abilities may be declining. Ask any
older adult and they will likely tell you
that they feel young: They report having just a few more aches and pains
and just a little more forgetfulness,
but not enough to affect their day-today functioning. To design systems
that respect this I still feel young belief, the technology needs to support
the users autonomy by empowering
users to adapt their routines in the
way they see fit rather than imposing
a rigid regimen that makes users feel
like they have a disability. Technology
for older adults should be designed to
be convenient and helpful rather than
rigid and stigmatizing.
A good example of technology that
preserves the users autonomy is the
spell checker found in word processing programs. Spell checkers only
intervene when the user has spelled
something incorrectly, and when they
do intervene, they give the option to
the user to accept or reject the spelling suggestion. Almost everyone accidentally mistypes or misspells words,
especially when typing quickly. Spell
checkers merely make typing and
spelling correctly more convenient
rather than making users feel badly
for being bad spellers or typists. Word
processors could enforce a rule that
imposes only correctly spelled words,
encroaching upon the autonomy of
the user, but they do not. Instead, spell
checkers step in only when the user has
made a mistake, and they conveniently
give the user a chance to correct it by
exercising his/her judgment about
how to spell the word. Similarly with
health behavior change, supporting
Ongoing frequent
feedback seems
to be instrumental
for maintaining
the improvements
in medication-taking
behavior.
the individuals autonomy is critical

because self-initiated behaviors driven
by the users intentions are generally
more effective and more sustainable
than behaviors driven by an external
agent. Individuals should be allowed
to set reasonable goals for themselves
(e.g., I want to take my medications on
time at least five times a week.) and be
supported with feedback technology
to make themselves aware of the progress toward their goals and to troubleshoot their behaviors.
Routines help us carry out regular
tasks with minimal demands on our
attention and effort. With this automaticity comes the occasional risk
of substituting the performance of a
task with the mere thought of carrying out the routine, as is often the case
with medication taking. Sensing and
feedback technologies in the home
can unobtrusively monitor the actions
of individuals and provide individuals with timely feedback about their
behaviors to support the mundane
routines important for maintaining
their health and independence. These
technologies must provide the right
amount of information and deliver
it at the right time without requiring
too much attention from the user or
imposing a rigid routine that compromises the individuals autonomy.
dwellSense is an example of an unobtrusive sensing system that provides
glanceable feedback to help individuals self-regulate, adapt, and improve
their medication taking routines
as demonstrated in a longitudinal,
10-month study. The same sensing
and feedback technique can also be
applied to other behaviors important

for maintaining independence. Monitoring other cognitive-related tasks
in the home such as dialing the telephone, multi-step meal preparation,
or operating new electronic devices
like a TV can also provide feedback
to individuals and their care network
about health information, such as
changes in vision or cognitive abilities. Smart glucometers can monitor how often people with diabetes
measure their blood sugar, support
their awareness of how often they
take measurements, and also provide
some assistance in interpreting the
values and identifying the causes for
deviations. Cars, equipped with GPS,
distance sensors, and other driveby-wire sensors can detect when an
individuals reaction time might be
slowing and provide feedback for individuals to drive more safely. Future
research should look into addressing
these challenges while considering
the task-supportive effects of realtime feedback while also considering
the impacts of long-term feedback.
Moreover, it should apply technologies that support individual autonomy to self-initiate and perform the
behaviors important for healthy, independent living.
References
[1]
Kripalani, S. Interventions to enhance medication

adherence in chronic medical conditions. Archives of
Internal Medicine 167, 6 (2007), 540.
[2] McDonnell, P.J. and Jacobs, M.R. Hospital

admissions resulting from preventable adverse drug
reactions. The Annals of Pharmacotherapy 36, 9
(2002), 13311336.
[3] Lee, Matthew L. and Dey, Anind K. Real-time
feedback for improving medication taking. In
Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems (CHI 14) (Toronto,
April 26May 1). ACM, New York, 2014, 22592268.
[4]
Bandura, A. Social cognitive theory of selfregulation. Organizational Behavior and Human

Decision Processes 50, 2 (1991), 248287.
[5] Lee, Matthew L. and Dey, Anind K. Reflecting on pills

and phone use: supporting awareness of functional
abilities for older adults. In Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems
(CHI 11) (Vancouver, May 712). ACM, New York,
2011, 20952104.
Biography
Matthew L. Lee is a researcher at Philips Research North
America. He is interested in bringing about the future
of mobile and ubiquitous computing by tracking and
making sense of the simple actions that people do in
their everyday lives to reveal rich patterns about peoples
behaviors and help people achieve their goals.
2014 Copyright held by Owner(s)/Author(s).

Publication rights licensed to ACM.
1528-4972/14/12 $15.00
43
feature
Seeing
Is Believing
Why visualization will play a critical role in bringing
big data decision making to a hospital bed near you.
By Megan Monroe
DOI: 10.1145/2676576
ts 2 a.m. in the Neonatal Intensive Care Unit (NICU) of a downtown hospital when an
alarm pierces the night. A bleary-eyed resident rushes in to investigate, weaving his
way to a computer that monitors the vital signs of each infant. The screen is adamantly
flashing an instruction to administer antibiotics to one of the babies. This is a daunting
request. For those of you who dont speak baby medicine, administering antibiotics to
an infant is not the same as popping a couple extra aspirin on a Sunday morning. You are
altering the bacterial landscape of an
immune system that is still developing,
which can result in all sorts of bad
down the road. In this case, however,
the computer has determined it is worth
the risk. But why? The vitals appear to
be normal. There is no visible cause for
concern. So the question is, should the
resident follow this instruction blindly,
without explanation?
This scenario, by the way, is taking
place in the near future. Today, there
would be no alarm until the infant
showed an obvious sign of distress,
such as spiking a fever. In this case, the
resident would immediately administer antibiotics, but it would probably be
too late to make a difference. It is also
worth considering the distant future,
in which the computer administers the
44
antibiotics as it pleases, and the resident isnt a resident at all because hes
finishing up a Ph.D. in bioinformatics.
That distant future is hardly some
singularity-dependent pipe dream.
Current pattern detection algorithms
can isolate trends that no human could
hope to unearth by hand. If a spiking
fever in the NICU is a sure sign of distress, but the symptom doesnt present
itself early enough for an effective intervention, these algorithms can scan the
data for invisible patterns that precede
the fever and suggest a course of action.
Before we break out the champagne,
however, there is one key thing to realize: These algorithms dont actually
produce answers, they produce probabilities. To a computer, a decision
is simply a probability paired with a
threshold. This can be problematic for

doctors, who are trained in biology,
chemistry, psychology, but not necessarily in statistics. Probabilities are not
their lingua franca, especially at 2 a.m.
The point is this: In order to arrive
at a distant future in which computers
are entrusted with critical, potentially
life-or-death decisions, we need to create a near future in which humans and
computers can arrive at data-driven
decisions together. If these machinelearning algorithms really want to
prove their worth, they need to be able
to learn directly from humans, not just
from the data trail we leave behind.
This means we need to find a way for
humans, across a wide range of backgrounds and expertise, to effectively
communicate with their data-crunchX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
ing counterparts. And while there is

much debate as to how this communication will ultimately take shape, there
is general agreement that a large component of it will involve visualization.
My graduate work at the University of
Marylands Human-Computer Interaction Lab was spent on the development
of EventFlow [1], a visualization tool that
allows users to assess event patterns
in the context of a large population of
patients (see Figure 1). This is done by
grouping together patients with the
same event pattern, and shifting the
most common patterns to the top of the
display. It allows researchers to quickly
answer population-level questions such
as Where were patients typically transferred after the emergency room? and
How often are patients re-admitted
to the ICU? Throughout the design
and development of the software, I collaborated with nearly a dozen medical
research teams on long-term, multi-dimensional data analytics projects. This
article draws on my experiences with
these EventFlow case studies, as well as
broader lessons learned outside of the
medical domain, to highlight three reasons why visualization will play a critical role in bridging the gap to a more
automated future in health analytics.
REASON NO. 1: THERE IS SUCH

A THING AS A STUPID ANSWER
What is Toronto? This answer,
chirped by IBMs supercomputer Watson in response to a televised Jeopardy question about U.S. cities, was
seriously stupid. The word stupid also
comes to mind when you read about
Target, an American discount retailer,
and their algorithmic decision to send
pregnancy coupons to a teenagers
home before she had even told her
parents that she was expecting. A computer will never skip an instruction or
calculate a percentage wrong, but when
they do err, the result can be obviously
and laughably stupid.
It is worth pointing out both of the
aforementioned stupidities were minor blemishes on the face of indisputable victories. Watson had dominated
Jeopardy so absolutely that one of the
contestants had resorted to taking per-
sonal jabs at it, and Target has made

untold millions using its predictive
marketing strategies. These isolated
missteps are a natural consequence of
probabilities, casualties of a low and
pre-calculated error rate. And yet still,
an average human could flag them as
incorrect in a single glance.
The blunders of artificial intelligence, however, become a lot less comical when they involve actual casualties.
What if the only reason the vital-signmonitoring computer in the NICU is
recommending antibiotics is because
the infant in question is in bed No. 4,
and the last three infants who spiked a
fever also happened to reside in bed No.
4? Perhaps bed No. 4 is indeed cursed,
but it doesnt take a certified doctor to
realize this logic seems flawed; it takes
a human. But in order for this common
sense smell test to take place, the computer must provide some form of explanation for the decisions its outputting.
Even if these oddities are extremely
infrequent, no one is going to underwrite an error rate that involves cursed
cradles and dead babies.
45
feature
The good news here is even a basic
visualization tool can typically reveal
answers that are rooted in stupidity.
For example, one of our EventFlow case
studies involved a dataset of the events
that precede and follow a surgery. The
researchers had previously been using
command-based query tools to explore
the dataset, which had been producing results that, again, seemed flawed.
Most notably, the calculation for the
average duration of a surgery seemed
slightly askew. When the dataset was
loaded into EventFlow, however, and
each patient record was displayed in
sequence, it became immediately clear
the calculation was being thrown off by
surgeries that spanned midnight. The
date of each patients surgery had been
recorded in a separate column from the
individual event timestamps, so events
that took place after midnight were appearing at the start of the day. While
these surgeries were rare, the data
anomaly was extreme enough to affect
aggregate calculations. The simple addition of a date to every timestamp immediately corrected the problem.
This example is not meant to highlight an overt stupidity on the part of
the humans or the computer, but rather how the disconnect between human
logic and computer logic can result in
stupid. The fact that a surgical team
cannot pause halfway through a procedure, travel back in time, and continue
the surgery at the beginning of the
day is so painfully obvious to humans
that it is easy to overlook this explicit
clarification. But from the computers
perspective, based on the data and the
instructions that it was given, that is exactly what happened. Why not?
Visual representations can serve as
a potent common language between
these two logics. They offer our best
hope at explaining the inner workings
of computational processes to professionals outside of the technical domain. Not only can visualization highlight overt logical inconsistencies, but
as well see next, they can also address
more subtle errors that arise from the
natural proliferation of new technology.
Reason No. 2: Infinity and BeyondPredicting the behavior of a process at
scale can be extremely difficult. The
financial industry learned this lesson
the hard way in 2008, when a complex
46
and ultimately misunderstood copula

function was used to price hundreds
of billions of dollars worth of mortgage
risk, a gross overextension of the functions original intention. Wireds 2009
article on the ensuing financial crash
concluded the following: One reason
was that the outputs [of the copula
function] came from black box computer models and were hard to subject
to a commonsense smell test. Another
was that the quants, who should have
been more aware of the copulas weaknesses, werent the ones making the
big asset-allocation decisions. Their
managers, who made the actual calls,
lacked the math skills to understand
what the models were doing or how
they worked [2].
If you swap out some of the nouns
in that previous quote, it can start to
sound eerily similar to our curious incident of the NICU resident. When a
solution works, the immediate temptation is to deploy it at scale. Computers will always be up for this challenge.
To them, one data point is the same as
one million data points, regardless of
whether the logic scales as gracefully
as the dataset. Ironically, it is the humans, with our comparatively limited memory and processing power,
who must decide how well the logic
will scale across hundreds of billions
of computations. This is a steep challenge, particularly when uncertainty
is involved (uncertainty is always involved). And the price of getting it
wrong can be globally severe.
Visualization can address the
difficulties of scale by abstracting
non-critical information out of the
display. In fact, this capability is
We need to find a
way for humans,
across a wide range
of backgrounds
and expertise,
to effectively
communicate with
their data-crunching
counterparts.
what prompted the very existence

of EventFlow. Medical researchers
were having trouble analyzing common event sequences within a large
dataset of patients. They had to
scroll through page after page of patient records; although this allowed
them to easily understand the individual attributes of any one record,
it made it impossible to glean insights about the dataset as a whole.
Our goal with EventFlow was to provide these researchers with a single
screen on which they could see every
event sequence in their dataset. To
do this, details about the time lapses between events were not included
in the primary display. This allowed
the visualization to group patients
by event sequence alone, creating
a much more compact display that
better supported a dataset-level understanding of event sequences.
REASON NO. 3: THE BLIND SPOT

Weve established computer logic is
imperfect and visualization can help reveal these imperfections, but we havent
touched on the fact that these visualizations can be interactive. Not only can
they help humans understand complex
algorithmic computations across large
datasets, they also provide an interface
through which humans can contribute
new information and feedback to the
machine. This contribution will be critical because, even when computer logic
is functioning exactly as intended, it is
always functioning within a bounded
corpus of information.
One of the most interesting EventFlow case studies came from the trauma bay at a childrens hospital. Nurses
and physicians in the trauma bay are
required to adhere to the Advanced
Trauma Life Support (ATLS) protocol,
a series of five steps that must be completed in a specific order during every
trauma resuscitation. The researchers with whom we collaborated were
interested in identifying common
deviations from the protocol in hopes
of providing feedback to the trauma
team as well as making systemic adjustments to better support adherence. In one of the most meticulous
data collections I observed, videos of
more than 200 traumas were reviewed
by hand, and each step of the ATLS
protocol was recorded at the time it

was executed.
When this data was visualized
in EventFlow, two unexpected findings became immediately clear. The
first was adherence to the ATLS protocol was far lower than expected,
suggesting this protocol does not
have as critical an impact on patient
outcome as initially assumed. The
second finding was instead of one
or two common deviations from the
protocol, the dataset contained 29
unique deviations, making it impossible to codify the results into a single
action plan for improvement. More
importantly, these findings shifted
the foremost research question from
how can we better adhere to this protocol? to which components of this
protocol are actually essential? The
researchers wanted to know whether
certain deviations were more predictive of negative outcomes than others. The problem was there was no
way to explore this new direction. Patient outcomes had not been recorded
because the data collection had been
designed under the assumption that
the correlation between outcome and
ATLS adherence was a given.
Compiling a perfect dataset is like
trying to catch a stream of water in your
hands. Some data is inherently inaccurate, some data is not available for logistical or legal reasons, and some data
simply isnt included because it didnt
seem necessary at the point of design.
Systems must be built to support a data
landscape that is always in flux, and a
large component of that will be seamlessly combining analytics from predetermined data sources with in situ human feedback.
Lets return to our example of the
resident in the NICU. What if, instead
of being dictated a blind instruction,
our resident arrives at the computer
to find a timeline of events and attributes that have been mined from
the infants vital signs? The computer
has found a particular pattern that
has preceded a fever in more than 95
percent of other NICU cases. This information has been perfectly visualized, making it extremely easy for the
resident to understand how the computer arrived at its decision. However,
maybe the resident knows this parX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
Figure 1. EventFlow consists of three panels: The control panel and legend (left),
the aggregated view (center), and the individual view (right).
Datasets like this one, which tracks patient transfers and outcomes, are typically
displayed as a vertical list of records, as seen in the individual view. Scrolling is
required to see the entire dataset, making it difficult to make holistic assessments. In
the aggregated view, common event patterns rise to the top of the display (1), and we
can immediately see anomalous errors (2) and outliers (3).
ticular infant had an allergic reaction

to the lotion a nurse was wearing earlier that day. If those symptoms are
appearing on the timeline, but have
already been resolved, the resident
might ask the computer to factor out
those particular events and recompute its findings. These changes may
reduce the risk of fever to only 35 percent, causing the system to rescind its
initial recommendation.
If the medical domain is looking
for inspiration on this front, it can
take cues from the self-driving car
industry. This industry has already
embraced the fact that real solutions
can come from imperfect data by incorporating human feedback. For example, if every car on the road were
networked, collisions could be avoided through car-to-car communication
and driving routes could be optimized
at a massive scale. Humans could happily sleep through their morning commute. Ten years down the road (pun
intended), this might be realistic, but
right now this data is simply not available. As a result, self-driving cars, in
their initial incarnations, will require
some degree of engagement from
their human drivers. The cumulative
task of driving will be a collaboration
between man and machine.
IN SUMMARY
One thing is certain: The answer to our
original question is no, our resident, or
any other medical professional, should
not blindly execute an instruction
without explanation or clarification.
Regardless of how much a computer
contributes to a medical decision, it is
the human who will ultimately be held
accountable for the decision-making
process. If computers hope to make a
contribution at all, they must find a way
to integrate their input into the humanthought process. And if these machines
possess true intelligence, they will do
this by leveraging one of the most powerful and mysterious components of
our human brains: the visual cortex.
References
[1]
EventFlow: Visual Analysis of Temporal Event

Sequences and Advanced Strategies for Healthcare
Discovery. Human-Computer Interaction Lab (HCIL),
University of Maryland.
[2] Salmon, F. Recipe for disaster: The formula that

killed Wall Street. Wired February 23, 2009.
Biography
Megan Monroe completed her Ph.D. in computer science
at the University of Maryland, and is currently working for
IBM Research in Cambridge, MA. Her research interests
include data analytics, visualization, and college football.
2014 ACM 1528-4972/14/12 $15.00
47
feature
Wearable Technologies:
One step closer to gait
rehabilitation in
Parkinsons patients
Wearable computing has the potential to fundamentally alter healthcare by
enabling long-term patient monitoring and rehabilitation outside of the lab.
By Sinziana Mazilu and Gerhard Trster
DOI: 10.1145/2676578
earable technologies have begun to take root in our everyday life. On-body
sensors are used to quantify how often and how well we perform our favorite
sport, while our smartphones have become our personal assistants
monitoring how active we are, our daily routines, the places that we visit, and
the activities we do. Wearable systems are already being developed for bipolar disorder,
cerebral palsy, chronic pain, and Parkinsons disease. The latter ranks among the most
common neurological disorders, with an estimated 7 to 10 million people worldwide
living with Parkinsons disease.
The symptoms of Parkinsons
disease include tremors, slow movements, rigidity of the limbs, shuffling
gait, and, in advanced stages, the freezing of gait. Freezing of gait is a sudden
inability to move, people with Parkinsons describe it as the feeling as if the
legs are glued to the ground [1]. Of
all the symptoms of Parkinsons, gait
freezing is the most feared, being the
main cause of falls and mortality in
people with the disease.
Although widespread, the causes
of Parkinsons disease remain unknown and there is currently no cure.
Existing treatments can only allevi-
48
ate symptoms through medication or

deep-brain stimulation with a brain
pacemaker. Freezing of gait is often
treatment-resistant. However recent
clinical studies have shown people
with Parkinsons disease respond to
rhythmic external stimuli [2], such as
metronome beats, which might help to
alleviate gait freeze. There is a simple
explanation: The human gait is rhythmic, and humans respond to rhythmic tones in everyday life, for example
when listening to music while walking
or jogging. However, the impact of a
continuously rhythmic pattern on the
Parkinsonian gait wears off with time,
in the same way as we tend to ignore

the rhythm of a song that we have listened to too often.
THE WEARABLE ASSISTANT

In the Wearable Computing Lab at
ETH Zrich we have been researching
the use of wearable sensors and smartphones to provide rhythmic auditory
stimuli to people with Parkinsons disease, helping them to alleviate or even
overcome freeze episodes.
Rhythmic stimulation combined
with gait-training exercises is a promising practice toward gait rehabilitation in Parkinsons disease [3]. A wearX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
Image by Ocskay Bence
able system will allow people to follow

these treatments in their homes, without any clinical supervision. Gait-training experts and clinicians are replaced
with the wearable system, which can
also act as a walking assistant during
daily-life activities.
There are two main research questions that underpin the development
of such a system: What kind of sensors
and what methods do we need to best
capture the gait-anomalies characteristics in Parkinsons disease? Where will
the wearable components be placed on
the patients body and how will future
users accept wearables?
Our answer to these questions is GaiX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
tAssist, a wearable gait assistant, which

supports rehabilitation exercises in Parkinsons disease [4]. The development of
GaitAssist threw us into the adventurous intersection of wearable systems,
machine learning, and human-computer interaction for healthcare.
GaitAssist uses on-body sensors
and a smartphone to detect gait freezing in real-time. Upon a gait freeze or a
similar gait pattern, a rhythmic auditory cueing is given to the user, supporting the subject in regaining the motor
function and to resume walking.
GaitAssist consists of up to two inertial sensors mounted on each ankle,
which send real-time data readings
to a smartphone. On the smartphone

an Android application analyzes the
sensing data and detects in real-time
whether the subject is experiencing a
gait freeze. If a freezing of gait is detected, a rhythmic metronome ticking
sound is started, which is synchronized with the rhythm of the users
gait. The rhythmic sound continues
until the user resumes their normal
walking pattern. A detailed description of GaitAssist and its components
is presented in Figure 1.
SENSOR TYPES
AND ON-BODY POSITION
Following previous research cues, we
49
feature
Figure 1: Eighteen people with Parkinsons disease and five healthy subjects were
asked to perform walking protocols in a laboratory setting while wearing nine onbody inertial measurement units (IMUs) and a smartphone.
Back sensor
Hand sensor
Thigh sensor
Smartphone
Ankle sensor
Foot sensor
Two of the IMUs were placed on the feet, two on the ankles, two on the thighs, two on
the arms, and one on the lower back of the subject. The smartphone was placed in a
trouser pocket.
Figure 2: The GaitAssist system consists of up to two wearable sensors.
(b)
(a)
(c)
(a) Attached on the ankle of the users, sensors are attached using specially designed
Velcro straps. Data sample from the wearable IMUs is sent in real time to a Samsung
S3 Galaxy phone (b) that acts as a wearable computer. Sensing data is analyzed in
order to detect the gait-freezing episodes. Upon motor block, a rhythmic sound is
provided for a limited period of time. The subject can choose to use a single earbud
(c) to hear the rhythmic biofeedback given by the system.
50
decided to use wearable inertial measurement units (IMUs) to capture the

characteristics of gait in order to detect
the motor freezing episodes.
On-body IMUs containing an accelerometer, a gyroscope, and a magnetometer are useful in capturing human
motion. They can be used to detect the
activity the user is performing [5] e.g.
sitting, walking, cycling, or running
and even offer precise-enough information to determine the correctness of a
movement, allowing (for instance) running skills to be assessed.
We asked five healthy participants
and 18 people with Parkinsons disease
and freezing of gait to each perform a
single three hour walking session in a
laboratory setting, over a period of three
weeks. The walking tasks were specifically designed by clinicians to induce
gait freeze in Parkinsons disease subjects and included turns, u-turns, and
walking in narrow, crowded corridors
or with varying cognitive load. During
the walking protocol, the participants
wore a system composed from nine inertial measurement units and a smartphone. The placement of the wearable
sensors is shown in Figure 2.
Each subject wore the system continuously for a period of two to three
hours, and sensor data was collected as
they completed the tasks in the walking protocol. Subjects then completed
questionnaires and participated in
open-ended discussions with clinicians, regarding the components that
the wearable system should have, and
where those components should be
placed on the body, in terms of comfort
and wearability. This co-design process concluded participants perceived
foot and ankle positions to be the most
comfortable for wearing on-body sensors. The smartphone in the pocket
was not noted at all, being already accepted as an everyday wearable tool.
Independent of the co-design study,
we analyzed the sensor data collected
during the walking sessions to check
which body position gave the best information to distinguish the motor blocking episodes from the other walking
events. Clinicians labeled all the freezing of gait episodes from the videos taken during the data collection, and these
labels were then synchronized with the
sensor data. Applying the same detecX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
Figure 3: A data sequence from an ankle IMUs accelerometer, gyroscope, and magnetometer from a subject with Parkinsons
disease
Data contains a gait-freezing episode and diverse walking events, such as getting up, turning, or stopping. We can easily observe
raw sensor data during the gait-freeze episode is similar to other gait events, such as sitting, starting to walk, or slowly turning.
3Axis Accelerometer Raw Data
freezing of gait
40
20
0
20
sitting
0
walking
turning 360 degrees

5
turning 360 degrees

10
15
walking
20
turn
walking
sitting
25
30
35
25
30
35
25
30
35
3Axis Gyroscope Raw Data

4
2
0
2
4
0
10
15
20
3Axis Magnetometer Raw Data

400
200
0
200
400
0
10
tion algorithm to IMU data from five different body positions showed the ankle
was the most informative position to
extract the properties of freezing events.
HOW TO DETECT FREEZING OF GAIT

FROM WEARABLE SENSOR DATA
Figure 3 shows a data sequence from an
ankle IMUs accelerometer, gyroscope,
and magnetometer from a subject with
Parkinsons disease. This sequence contains a gait-freeze episode and diverse
walking events such as getting up, turning, and stopping. We can easily observe
that raw data during the gait-freezing
episode is similar to the data collected
during other gait events, such as sitting,
starting to walk, or slowly turning.
In our problem it is just as important
to avoid false gait-freeze detections, as
it is to detect the freezing episodes, e.g.,
to incorrectly classify a stand up event
as a gait-freezing episode, only because
their IMU data patterns share similarities. Thus a very important part of our
work was to find the best features in
15
20
the IMU data to characterize the motor

block events and to develop methods to
detect such episodes in real-time.
Fast Fourier Transformation (FFT)based features from acceleration are
used to characterize and quantify the
gait. We followed this direction and
compute four gait-freeze specific FFT
features in a sliding window manner
from the raw acceleration data. Figure
4 contains details about FFT-based
features and an example of how they
change during normal walking and
freezing of gait. We extended our feature extraction to statistical features,
e.g., mean, standard deviation, min,
and max from accelerometer, gyroscope, and magnetometer data. But the
most informative features for our data
are the FFT-based ones extracted from
acceleration data.
The gait-freeze detection algorithms
consist of decision trees, trained on the
acceleration data from the ankle position, collected from the 18 Parkinsons
disease subjects. Machine learning
models and decision trees in particular

are robust in terms of both detection
performances and computation time
for human activity recognition and gaitfreeze detection. The feature computation time and the classifier decision time
are very important in detecting risky
gait anomalies such as freezing of gait.
A freeze episode needs to be detected as
soon as it starts, to avoid the onset of a
total motor block or a fall, and to help in
resuming walking as early as possible.
CLOSING THE LOOP: A PRELIMINARY

EVALUATION OF GAITASSIST
The best evaluation a healthcare wearable system can have is to pass acceptance tests with its users. To test
GaitAssist, we asked five additional
subjects with Parkinsons disease and
freezing of gait to set up and wear GaitAssist during a gait-training protocol
in a hospital setting. The subjects had
not participated in the previous design
and development process of the system, and had not used a wearable sys51
feature
tem in other circumstances before the
evaluation protocol started.
Subjects were asked to perform a
gait-training and rehabilitation protocol assisted by the wearable system, as
shown in Figure 5. The protocol included sessions of walking designed to provoke gait freeze events, such as u-turns,
360-degrees turns, or sit-to-stand tasks,
but also sessions of natural walking in
crowded and narrow corridors, stairs,
and elevators of the hospital. Each session lasted around 30 minutes and subjects were asked to repeat the protocol
on three different days. The protocol was
video recorded and synchronized with
the GaitAssist sensing data, and as in the
previous experiment clinicians detected
102 gait-freezing events from the videos.
The first test: How well does the
system function? GaitAssist successfully detected 99 gait-freezing episodes and started auditory cueing in
response, typically with latency smaller than 0.5 seconds after the start of
a gait-freeze event. The three missed
gait-freezing events were shorter than
0.5 seconds, and therefore difficult to
detect; 57 false alarms occurred in total during the study, meaning GaitAs-
sist started rhythmic auditory cueing

at moments without gait freeze. This
typically happened during unusual
and anomalous gait motions that resembled freezing, such as turning
with very small steps, sudden stops,
or during sit-to-stand and stand-to-sit
movements. The latter triggers were
due to the lack of these types of movements in the training data.
The total time required to compute
the FFT features from the sensing
data,and for the gait-freeze detection
algorithm to make a decision, is at most
six milliseconds, measured in real-time
on the Samsung Galaxy S3 mini phone.
Regarding resource consumption, in a
realistic usage setting GaitAssist consumes less than 1 percent of the battery
power per hour. However, the IMU batteries last for at most four hours, which
is sufficient to use the system for the
gait-training exercises but limits the duration for which the system can be used
as an assistive device.
The second test: Is the system acceptable to users? At the end of each
testing day we asked the five subjects
to complete a feedback questionnaire
regarding the usability, feasibility, and
Acceleration [m/s2]
Figure 4: Twenty seconds of walking and turns, including a gait-freezing episode.

40
freezing of gait
Acceleration magnitudes [raw data]
30
20
10
0
Power freezing of gait band [38] Hz
10
Power
10
12
14
Power locomotion band [03] Hz
16
18
20
Total power [08] Hz
10
10
12
14
16
18
20
12
14
16
18
20
Freeze index
10
Freze index feature
5
Time [s]
We compute the raw-acceleration magnitude and compute 4 FFT-based features

in a sliding-window manner: Power on the freezing of gait band [38] Hz, power on
the locomotion band [0.53] Hz, the total power in the [08] Hz band, and the freeze
index, which is the ratio between the power on the freezing of gait band and the
power on the locomotion band. Power-on freeze band and freeze index features
increase prior to or during the gait-freeze compared with the walking periods, while
power on the locomotion band (PL) values decrease before or during freeze. While
the locomotion-power feature is similar in cases of turning, standing, sudden stops,
and freezing of gait, the total power feature helps in distinguishing between turns
and freezing of gait, for example. The combination of the four features can give an
accurate description of the freezing-of-gait episodes.
52
comfort of the system. The usability

part assessed the participants experiences with the rhythmic audio feedback during freeze episodes. The feasibility part inquired about the possible
use of the GaitAssist system at home,
and the participants motivation to use
the system in daily life. In the final part
of the questionnaire, we addressed the
usage of the system from a technical
point of view, and the user-friendliness
of the software and sensor setup. Additionally, we performed informal discussions between a clinician and the
user to gather more detailed feedback.
Participants reported they were
satisfied with the wearability and performance of GaitAssist. They enjoyed
using the system and felt it was reliable and accurate. The sensor attachments did not trouble the users during
walking, although they expressed the
desire to wear the sensors under their
trousers. Subjects reported they often
forgot they were wearing the sensors
during the trial.
They perceived the audio feedback
played through earphones as easy to
listen to, even in noisy environments
They also reported the auditory feedback was always triggered at the right
time when a long freezing of gait episode occurred, and with low response
latency, i.e., 0.5 seconds, which corresponds to the latency we measured.
All participants felt they benefited
from the support of GaitAssist during
freezing episodes. Figure 6 shows their
perception on whether the rhythmic
auditory feedback provided by GaitAssist could reduce the number of freezing events and whether it could reduce
the duration of those episodes. They reported the freeze duration was shorter
compared to their normal experience,
but using the system would probably
not decrease the number of such episodes. Although rhythmic stimulation
was provided in some false-positive
cases, i.e. without a motor block in
progress, subjects reported this might
in fact have prevented them from experiencing a gait freeze, as they felt they
were having difficulties walking in
some of these instances.
INSTEAD OF CONCLUSIONS
At the end of the day, nothing is more
rewarding for researchers than seeing
Figure 5: The GaitAssist system during in-the-lab validation with five Parkinsons
disease subjects.
From left: A subject wearing the wearable sensors during the protocol, a subject
performing a figure-eight-like walking task, and another subject preparing to perform
a protocol session, which includes sit-to-stand and turning tasks.
the positive impact of their developed

system on its target users. GaitAssist
was well received and accepted by its
users, who stated they dont feel alone,
and the system is like someone who
supports them, they could wear it all
day long, as it helps with decreasing the
gait freeze severity, and it is very useful and improves the gait.
Following these positive results and
users feedback, clinical researchers
are running an in-home trial with more
than 20 subjects with Parkinsons disease and freezing of gait to study whether GaitAssist can help improve their
gait and alleviate gait freeze in the longterm. Users are performing specially
designed gait-training exercises without
any clinical supervision, assisted only by
the wearable system. The hope is after a
few weeks of such training, users will
have improved overall gait quality.
GaitAssist is designed to be an as-
sistant for its users and even to replace

the clinicians help outside the lab,
opening the opportunity for people
with Parkinsons disease to perform
gait-training protocols in their homes.
Up to now such treatments were given
in clinical settings or under the supervision of a physiotherapista burden
for the healthcare system in terms of
costs and personnel.
In this article weve focused mainly
on one function of the wearable systemthe real-time detection of gaitfreezing episodes and the start of
rhythmic auditory stimulation upon
these eventsbut the wearable system
also supports gait-training exercises
designed by physiotherapists. It has
a simple to use interface allowing the
user to choose the type of training and
exercises to be performed.
The system is connected to a telemedicine server and sends statistics
Figure 6: The five participants individual scores to the following statements:

(a) GaitAssist helps in reducing the freezing of gait number of episodes, and
(b) GaitAssist helps in reducing the freezing of gait durations.
Reduce the gait-freeze number Reduce the gait-freeze duration
Likert score
5
4
3
2
1
0
Subject1
Subject2
Subject3
Subject4
Subject5
Subjects were asked to give answers in a Likert score format, where 1 means they
strongly disagree with the statement and 5 means they strongly agree with the
statement. All of them appreciated that GaitAssist supports them in decreasing
the freezing of gait duration, but not the number of episodes.
and raw information about sensing

data and gait-freeze detection. Clinicians are able to remotely monitor the
training and progress of the patients,
and change the treatment and exercises setting accordingly, without requiring the user to visit the hospital or the
clinician to visit the users home.
The detection of gait freeze is a difficult yet very interesting problem to
solve, and one, which is still open-ended. But detection implies gait-freeze
events still occur. A different approach
would be to be able to predict periods
of gait difficulty, and to provide a rhythmic cueing or alerts, in order to help the
user to completely avoid these episodes.
Meanwhile, we hope wearable technology, in general, will enrich the
healthcare system, and GaitAssist will
help in building our knowledge of Parkinsons disease rehabilitation.
References
[1]
Nutt, J.G., Bloem, B.R., Giladi, N., Hallett, M., Horak,

F.B., and Nieuwboer, A. Freezing of gait: Moving
forward on a mysterious clinical phenomenon.
Lancet Neurology 10, 8 (2011), 734744.
[2] Donovan, S. Lim, C., Diaz, N. et al. Laserlight cues for

gait freezing in Parkinsons disease: An open-label
study. Parkinsonism & Related Disorders 17 (2011),
240245.
[3]
Espay, A., Baram, Y., Dwivedi, A. et al. At-home

training with closed-loop augmented-reality cueing
device for improving gait in patients with Parkinsons
disease. Journal of Rehabilitation Research &
Development 47 (2010), 573581.
[4]
Mazilu, S., Blanke, U., Hardegger, M., Trster, G.,

Gazit, E., and Hausdorff, J.M. GaitAssist: A daily-life
support and training system for parkinsons disease
patients with freezing of gait. In Proceedings of the
32nd annual ACM Conference on Human Factors in
Computing Systems. ACM Press, New York, 2014,
2531-2540.
[5] Bulling, A., Blanke, U., and Schiele, B. A tutorial on

human activity recognition using body-worn inertial
sensors. ACM Computing Surveys 46 (2014), 133.
Biographies
Sinziana Mazilu received her Dipl.-Ing. and M.Sc. degrees
in computer science and information technology from
Politehnica University of Bucharest in 2009 and 2011,
respectively. She joined the Wearable Computing Laboratory
at ETH Zrich as research assistant in 2011. Her research
interest lies at the intersection of wearable computing for
healthcare, human computer interaction, applied machine
learning, and context recognition from on-body sensors.
Gerhard Trster received his Dipl.-Ing. degree in electrical
engineering from Darmstadt and Karlsruhe in 1979
and his Dr.-Ing. degree from the Technical University
(Darmstadt, Germany) in 1984. He was involved in the
research on design methods of analog/digital systems in
CMOS and BiCMOS technology for eight years at Telefunken
Semiconductors. Since 1993, he has been a full professor
of electronics at ETH Zrich, heading the Electronics
Laboratory. At ETH, he established the multichip module
(MCM) electronic packaging group. In 2000, he founded the
Wearable Computing Laboratory, where he is involved in
interdisciplinary research combining IT, signal processing,
electronic platforms, wireless sensor networks, smart
textiles, and human-computer interaction.
2014 ACM 1528-4972/14/12 $15.00
53
profile
PROFILE DEPARTMENT EDITOR, ADRIAN SCOIC
Trevor van Mierlo

The Story of Building a Startup
in Health Informatics
DOI: 10.1145/2685368
Anyone whos been

a member of tech
culture for a while
can acknowledge the
startup phenomenon
has been responsible
for infusing even more technology into
our lives. However, despite the fact that
weve managed to move almost every
facet of our mundane existence online
from working to bondinghealthcare has
stubbornly resisted digitization attempts
over the years, making health-startups
today seem like an uncharted land of
uncertainty and untapped potential.
To find out what it takes to seize the
opportunity and build a successful
business in the risky environment of
healthcare startups, I went to the source.
Trevor van Mierlo lives in Canada.
He is the founder and CEO, as well as
acting CSO, of Evolution Health, which
specializes in building digital solutions
to improve patients adherence to
treatment and medication. The company
now has offices in both Toronto and
Silicon Valley, and it does business
with partners from all over the world.
According to van Mierlo, the backstory
is a long, inspiring journey of constant
learning and self-improvement.
PREPARING FOR A CAREER

IN HUMANITIES
I was born in a small town in Northern
Ontario, where I grew up infor lack of a
better terma family of proletarians; I
was actually the first person in my family
to go to university, van Mierlo recalled.
And to make his story truly surprising,
even though he is now the CEO of a
healthcare tech company, his career
started out with an undergraduate
degree in humanities, not a STEM field.
Actually, as a teenager, I just spent
most of my time playing hockey instead
54
of video games or doing some minor

programming on VIC-20s or Commodore
64s, so technology wasnt my forte, he
explained. He pointed out that as head
of a company with its own server farms,
developers, designers, and programmers,
he now needs to have an understanding of
all the languages and be able to work with
all the technologies involved.
My first degree was a double major
in English and history, and I really loved
school, and I was going to pursue my
doctorate when, quite frankly, one of my
advisors said to me that I was probably
making a bad choice to do a Ph.D. in
theory and criticism, because it was going
to cost me a lot of money, and it would
be very difficult for me to find a job, he
admitted. Since he was the type of person
who really loved data and really loved
academics, he was encouraged to go out
into the world and find a job in something
that was new and uniquesomething that
he loved because school would always
come back to him later.
LIFE AFTER GRADUATION

After graduation, van Mierlo took the
advice of his mentors and went on to
work for a health informatics startup
company called Mediconsult.com, of
which he was the ninth employee during
the dot-com bubble of the 1990s.
The company ended up creating
360 employees within a two-year time
period, so it was baptism by fire for
me, he recalled, adding the stock for
the publicly held company went from
about 49 cents to between $23 and
$24 in those two years. They company
had managed to raise (and spend) as
much as $80 million in its last round
of financing. It was at Mediconsult
that van Mierlo saw potential in the
healthcare industry, and began to lay the
foundations for his future work in health
informatics. In his own words, the late

90s were a really exciting period for
the dot-com boom, because people were
throwing money at anything you could
digitize. He was playing a technical, a
business development, and a product
management role in the company, which
allowed him to earn experience with
both the business side of pharma and
the research side of behavioral health. It
was also at Mediconsult that he first got
a taste of the power of data, which was
appealing to both the business people,
for whom it was providing return on
their investment, and to the behavioral
psychology researchers, for whom it was
providing new avenues for study.
Thus, when Mediconsult later went
out of business in the same way many
dot-com companies did, he recognized
the opportunity to go out into the
field and develop programs with fairly
specific return on investment criteria
for the different stakeholders he had
previously worked with. In 2000, he
founded his own consulting company
called Evolution Health.
As his work at Evolution Health was
taking off, van Mierlo was excited to see
his occupation merge into the academic
side of things when he began going
to international medical conferences
and presenting the outcome data. It
became apparent to him that he needed
a masters degree in health in order to
be accepted as an insider within the
international research community, so
he decided to go back to university.
He earned a masters of science and
community health, with a concentration
on addictions and mental health.
I think its a little bit of a story in
itself how when youre setting out in your
career, sometimes its good to follow an
academic path thats laid out for you.
Getting a masters of science, as I did,
Photo Credit TK
was not something that would have

naturally occurred to me after my first
degree, he confessed.
As things snowballed with Evolution
Health, and the business started
to grow, van Mierlo again realized
while talking to potential investors
and clients that he was still lacking
somethingthis time, it was business
speak. He was quick to take his own
advice, and enrolled in a program at the
University of Toronto called the Omnium
Global Executive MBA, which was a joint
program between the Rotman School
of Management and the University
of St. Gallen in Switzerland. He was
therefore able to go to class at different
universities in countries such as Brazil,
China, India, as well as Europe and
North America. This opportunity allowed
him to learn about doing business
internationally, in different cultures,
while at the same time getting the
financial, accounting, and supply chain
knowledge that he needed in order to
advance Evolution Health.
He didnt stop there, however. By
now, weve presented at almost 100
academic conferences; we have at
least 20 peer-reviewed publications,
so weve done a lot; and we work in
diverse areas, from smoking cessation,
to depression, panic disorder, colitis,
and arthritis, and more recently even
schizophrenia. The ability to develop a
methodology and be able to apply it to
different disease conditions and also
different cohorts of patients is really
fantastic, because we just keep learning
and developing new things every day,
van Mierlo enthusiastically explained.
He is currently pursuing a doctorate
degree with a focus on patient profiling
and the forecasting of health behaviors
through predictive algorithms, through
which he hopes to share his findings on
the demographic and psychographic

characteristic of superusers in health
support communities .
MANAGING A HEALTHCARE
STARTUP
While talking about the process of
founding and growing a healthcare
startup business, van Mierlo explained
the key challenge he had to face was
overcoming the systemic problem of
digital health, which is marrying three
disciplines that often dont get along
with each other: business, healthcare,
and technology.
On one hand, you have technology,
which moves very fast, on the other hand
you have healthcare, which moves very
slowly, and then you have the business
and financing side of things, which is
really focusing on maximizing profit.
Getting these three types of people in a
room is very interesting, because their
underlying philosophies are not the
same, and they often come from very
different perspectives.
He also warned being an entrepreneur
requires having a self-motivated
outlook on life, and should be fully
prepared to make more mistakes than
you have successeswhich is in itself
a very powerful life experience. To be
successful, he explained, one needs
to realize with an environment like
healthcare and digital health, everything
is progressing so fast if you dont remain
light on your feet and you become too
invested in a certain type of technology,
it can cost you everything.
Being able to adapt technology is
extremely important, because what
worked in 2010 no longer works in 2015,
and, from a business perspective, it
can become very expensive to invest
in things which are outdated quickly,
he added. According to van Mierlo,
Evolution Health had made the mistake

of being stuck with legacy in the past. He
explained agile development is crucial
for the success of a health informatics
startup, which is not always true for
large software companies,
A CLOSING WORD
While van Mierlo thinks e-health is going
to become as ubiquitous in the future
as social media is today, he does warn
that the path has not yet been cleared
despite investments of hundreds of
millions of dollars in digital health over
the past couple of years.
The only thing Id really say, he
concluded, is to keep an open mind.
When I was in school and I had my double
major in English and history, I would have
never believed you if you had told me
that I would end up having four masters
degrees, that I would have a number of
publications under my belt, and that I
would also be running a company thats
doing business in a number of countries.
However, the strength that I do have is
being able to recognize opportunities
and find holes in both academia and
business. So I think that if youre able to
go into both business and school with
an open mind, the places you can go are
quite amazing.
Copyright held by Owner(s)/Author(s).
55
end
A world map listing the latest known OpenMRS implementations and their intention of use.
LABZ
The Regenstrief Global

Health Informatics Group
Indianapolis, Indiana
Editors Note: In this issue Suranga
Nath Kasthurirathne talks about his
research and fellowship experience in
supporting and contributing to a number
of Open Source Health Informatics projects,
specifically the Open Health Information
Exchange (OpenHIE) projectSomdip Dey
s a student, my association
with the Regenstrief Global
Health Informatics (GHI)
group began in 2011, when
I was an intern for one of their open
source projects. Regenstrief GHI falls
under the domain of the Regenstrief
Institutes Center for Biomedical Informatics .The group is led by Dr. Paul
Biondich, who has been a key cham-
56
pion and architect of many of Regenstriefs global health initiatives. The

GHI team focuses specifically on addressing issues pertaining to global
health information technology, and
is widely known for its efforts to improve healthcare across some of the
most underserved populations on
earth. Regenstrief, in collaboration
with Partners In Health (Boston) and
the Medical Research Council (South
Africa), helped found the OpenMRS
project in 2004. OpenMRS is an open
source effort to build a framework
for electronic medical record (EMR)
systems in resource constrained environments. The rapid expansion and
acceptance of this project contributed
greatly toward the formal establishment of the GHI group, and helped
formulate many of the groups key
aims, values, and areas of focus.
The GHI group strives to:
Be a world leader in pragmatic
health information solutions.
Promote a global community
where the promise, outcome, and
real-world value of health information
technologies are recognized.
Ensure the environments we
serve are empowered to implement
and maintain this technology on their
own.
Directly contribute to the
strengthening of large health care
systems.
Currently, the main areas of focus
of the GHI group include: the Academic Model Providing Access To Healthcare (AMPATH), Open Medical Record
System (OpenMRS) and Open Health
Information Exchange (OpenHIE).
My internship was part of the
Google Summer of Code program. My
work on OpenMRS boosted my interX R D S W I N T E R 2 0 14 V O L . 2 1 N O . 2
Early Radiography photograph by Wilhelm Rntgen. Bone construction photograph by Zygorfi (https://commons.wikimedia.org/wiki/File:Bonereconstruction.jpg)
Real-time data collected on mentions of symptoms

on Facebook and Twitter allows sites like Sickweather.com
to plot a disease heatmap of the world.
est in health informatics, and encouraged me to continue contributing to

the project as a volunteer. After continuing to work on the project under
various roles, I went on to become a
mentor for other interns who were new
to OpenMRS.
In August 2013, I became a fellow of the Regenstrief GHI group. As
part of my fellowship, I was able to
continue working under the capable
guidance of Drs. Paul Biondich, Burke
Mamlin, and Shaun Grannis. And as
part of my commitment to OpenMRS,
I am heavily involved in mentoring
the next generation of budding health
informaticians. I also assist in the development of OpenHIE, focusing on
the evaluation and measurement of
clinical care delivered by the Rwanda
Health Information Exchangethe
first ever implementation of OpenHIE. As a graduate student, I feel privileged to see this project grow, and
have learned tremendously by watching, and contributing toward its design and development.
I also play a significant role in exploring the future of healthcare interoperability standards and their
application to both OpenMRS and
OpenHIE. I recently led an effort to
design and build support for Fast
Healthcare Interoperability Resources (FHIR)a widely accepted interoperability standard used to exchange
healthcare information between different hospital systems and clinics.
In addition, I am also involved
with other efforts to improve the efficiency and reliability of public health
registry reporting. We believe the automated diagnosis and reporting of
health conditions can help address
this limitation, and are working on
production worthy approaches to address these issues.
BACK
Radiography
Modern medical imaging encompasses a diverse set of techniques
used for scientific study, and non-invasive diagnosis of many medical
conditions. Many of you are probably familiar with the names of
several such techniques, and most of you have likely been subject to
at least one imaging technique at some point in you life. Perhaps the
most familiar category of techniques though is radiography, which
refers primarily to methods using X-rays.
The X-ray was discovered in 1895 by Professor Willhelm Rntgen.
During his experimentation with cathode rays, he observed his
equipment was producing some other invisible rays that were capable
of penetrating books on his desk. After studying the phenomenon in
greater detail, he produced the first X-ray photograph, radiographing
his wifes hand. The potential medical application was clear and
quickly put to use. Soon after this discovery, in 1896 John-Hall
Edwards became the first to use radiography in clinic and surgical
practice, while Thomas Edisons fluoroscope became the common
imaging device in use.
X-rays were not well understood in the very early days, however.
The image quality was lower and the dangers of radiation were not
immediately acknowledged. Since then radiography has come a long
way in the last century, with advances in X-ray emitters and digital
detection technology that allow for faster imaging at lower doses of
radiation. More recent developments in high-resolution radiography,
and computed tomography with fast 3-D reconstruction of scans, has
pushed the popularity of the technology even further. It continues to
expand into more applications, medical and otherwise.
Finn Kuusisto
Early Radiography
Modern Radiography
Timeperiod
1890s
Present
Image Dimensionality
2-D
2-D, 3-D
X-ray attenuation, phase shift,
backscatter
Digital detector
Radiation measurement X-ray attenuation

Biography
Suranga Nath Kasthurirathne is a second year Ph.D.
student in health informatics at the Indiana University Purdue University (IUPUI), Indianapolis. He is also a fellow
of the Regenstrief Global Health Informatics (GHI) group.
Detector Type
Photo plate or film
Uses
Medical
Medical, industrial, security,

microscopy
57
HELLO WORLD
The Anatomy of a Human

Disease Network
BY MARINKA ZITNIK
ools of network analysis have

recently been applied to many
complex systems, to both
simplify and highlight their
underlying structure and the relationships that they represent. The results
obtained from network-based approaches provide not only insight into
interactions between online users, but
also new clues about how to improve
our understanding of biological systems. Network medicine in particular,
a network-based approach to studying
human disease, has proven effective
in studying interdependence between
molecular components in cells, and
in identifying disease modules and
biological pathways [1].
The data we consider is usually
in the form of a set of records, each
of which describes one object in the
system. These objects may be
physical entities such as planets,
stars, or people, for example
customers; or may be abstract entities
such as diseases. Given a social or a
biological system, many kinds of data
can be collected. For example, a star
might be described by the strength
of its spectral emissions at particular
wavelengths; a customer may be
profiled using his or her demographic
data; and a disease might be described
with a set of genes whose mutations
are implicated in this disease. Data
about these systems are often
represented in the form of networks.
Network abstractions often turn out
to be practical, but remain highly
complex and are hence an active area
of research.[2, 3, 4, 5].
In this column, we explore the
human disease network [2] and
demonstrate how network-based
tools can help us understand relations
between diseases at a higher level
of organismal organization without
considering any prior biomedical
knowledge.
58
Figure 1: The human disease network. Every node corresponds to a distinct

disorder and is colored based on predicted network community to which it
belongs. The size of each node is proportional to the number of genes implicated in the corresponding disease. We labeled diseases associated with
many genes.
ON CONSTRUCTING
THE NETWORK
To construct the human disease
network we follow the influential work
of Goh et al. [2]. Nodes in the network
represent diseases and two diseases
are connected to each other if they
have at least one gene in common
whose mutations are associated
with both diseases. Disease data and
their associations with genes are
obtained from the Online Mendelian
Inheritance in Man (OMIM), which
is a comprehensive and regularly
updated online resource (http://www.
omim.org). It contains information
on all known Mendelian diseases and
tens of thousands of genes. Readily
prepared human disease network
data are available for download from
supplementary material [2]. The
reader may also access the dataset
by visiting the diseasome website
(http://diseasome.eu). Once one has
downloaded the network dataset,
the human disease network can be
constructed as shown in Listing 1. We
also use Igraph (http://igraph.org),
a network analysis package suitable
for explorative analysis of small and
medium-sized networks. Alternatively,
one may want to check SNAP (http://
snap.stanford.edu), a scalable graph
mining library that can handle massive
networks. We visualize our network
using Vis.js (http://visjs.org), which is a
dynamic, browser-based visualization
library allowing manipulation and
interaction with the data.
STRUCTURAL PROPERTIES
OF THE DISEASE NETWORK
We first examine some properties
of the human disease network as
computed by the script in Listing 2.
Of the many diseases in the OMIM
compendium (listed at the time of the
Goh et al. paper), 867 diseases have at
least one connection to other diseases
and there are 1,527 edges altogether.
If every disease would be independent
of others in terms of mutated disease
genes, then the network would
fall apart into many single nodes
corresponding to individual diseases
and small disconnected components of
few closely related diseases. However,
Listing 1: A Python script to construct the human disease network. The list
of diseases, genes, and associations between them was obtained from the
Online Mendelian Inheritance in Man and from Goh et al. [2].
from igraph import Graph
# reading weighted edge list from a file
f = open(disease.net.w.txt)
f.readline()
g = Graph.Read_Ncol(f, weights=True, names=True)
g = g.as_undirected(mode=collapse)
f.close()
# setting disease names as node attributes
f = open(supplementary_tableS1.txt)
f.readline()
f.readline()
did2name = dict([line.strip().replace(, ).split(\t)[:2] for line in f])
f.close()
g.vs[disorder] = [did2name[did] for did in g.vs[name]]
Listing 2: A Python script to calculate various structural properties of the

human disease network.
print Nodes: %d % g.vcount()
print Edges: %d % g.ecount()
cmp = g.components(mode=strong)
print Strong conn. comp: %d % len(cmp.sizes())
print Giant component: %d % cmp.giant().vcount()
print Closed triangles: %d % g.as_directed().triad_census().t300
print Diameter: %d % cmp.giant().diameter()
d = sorted([(n.degree(), n[disorder]) for n in g.vs], reverse=True)[:10]
print Max deg.:\n %s % \n .join(%d (%s) % (deg, dis) for deg, dis in d[:5])
print Clustering: %3.3f % g.transitivity_avglocal_undirected()
Listing 3: A Python script to find network communities with structure based

on random walks, label propagation method and the Infomap method.
# detecting communities in the HDN based on random walks
rndwlk = cmp.giant().community_walktrap(steps=3).as_clustering()
mmbr_rndwlk = rndwlk.membership
print Modularity (random walks): %3.3f % rndwlk.modularity
# detecting communities in the HDN with label propagation-based method
lblprp = cmp.giant().community_label_propagation()
mmbr_lblprp = lblprp.membership
print Modularity (label propagation): %3.3f % lblprp.modularity
# detecting communities in the HDN using the Infomap method
infomap = cmp.giant().community_infomap()
mmbr_infomap = infomap.membership
print Modularity (Infomap): %3.3f % infomap.modularity
print Codelength (Infomap): %3.3f bits/step % infomap.codelength
59
the largest connected component

contains 516 diseases, suggesting
there are many connections between
diseases that are related at both
a lower and higher level of cellular
organization. Whereas most diseases
are connected to only few other nodes,
there is a small number of highly
connected disease nodes known as
hubs, such as colorectal cancer (its
node degree is 50), breast cancer (it
has 30 links) and gastric cancer (it has
27 links).
Next, we are interested to see if the
network exhibits community structure.
Communities or modules are densely
connected subsets of nodes with only
sparse connections between them.
Many studies [2, 3, 4, 5] discovered
communities that corresponded to
functional or behavioral units within
networks, such as social groups in
social networks or protein modules
in biochemical networks. This implies
that we may be able to gain insights
into systems, whose operation is less
well understood, as is in the case of
human disease network, by detecting
and examining their communities. A
simple way to estimate the overall
presence of community structure in
a network is to calculate triad census
(Listing 2) and compare it to that
obtained from an appropriately defined
and randomly generated network. Triad
census classifies every triplet of nodes
to one of four possible types of triads
and for each type reports the number
of triads present in the network. A
triplet consists of three nodes that
are connected by either zero, one,
two, or three links, the latter is known
as a closed triad. The human disease
network has 1,517 closed triads
labeled as triplet type 300 [2]. We
also calculate clustering coefficient
of the network, which provides an
estimate of probability that two
neighbors of a node are connected.
More precisely, clustering coefficient
is a ratio of the number of closed
triads to the number of connected
triplets in the network. The human
disease network has a high average
clustering coefficient, which is 0.81.
For comparison, a recent dataset of
friends lists from Facebook has an
60
average clustering coefficient of 0.61

(http://snap.stanford.edu/data).
NETWORK COMMUNITY
DETECTION
The discovery of community structure
is a challenge of great interest, and
methods for community detection
have attracted considerable attention
across many disciplines [2, 3, 4, 5]. As
we have just seen, and as concluded by
many studies [1,2], there is a strong
indication that human disease network
contains communities of closely
interconnected diseases. We shall
consider here three approaches to
search for communities (Listing 3) in
our network.
Our first approach to organize
the disease network is based on
random walks [5]. The algorithm is
called Walktrap [5] and builds upon
the intuition that random walks on
a network tend to get trapped
into densely connected parts that
correspond to communities. The
Walktrap algorithm runs short random
walks (see steps parameter in Listing
3) to estimate similarities between
nodes and between communities, thus
defining a distance. Distance scores
are then used to iteratively merge the
nodes into communities and to obtain
a hierarchical community structure.
An induced hierarchy of network
partitions is then scored against
modularity, a quality function widely
used in many community detection
approaches, to select a partition that
captures well the community structure
of the data. Figure 1 shows partition of
the disease network as was detected
by Walktrap. Disease nodes are colored
based on communities to which they
belong and nodes associated with most
disease genes are labeled. One can
see Walktrap was able to successfully
group diseases and automatically
recognize a number of disease classes
related to cancer, hematological,
muscular, and ophthamological
diseases, among others.
Our second method finds
communities with a label propagation
algorithm [4]. The algorithm initializes
every node with a unique label and
then in an iterative manner reassigns
node labels in a way that each

node adopts the label that most of
its neighbors currently have. The
assumption is that by the end of this
iterative process densely connected
groups of nodes reach a consensus
on a unique label, hence defining
a community (Listing 3). Another
method we apply to the disease
network is Infomap [3]. This is a rather
different technique that attempts
to model information flow between
diseases. It uses probability flow
of random walks on a network as a
surrogate of information flow and
formulates a coding or compression
problem to efficiently describe random
walks. Infomap uses computational
search to find a partition into
network communities that minimizes
the expected description length
of a random walk. The particular
partitioning of the disease network
picked up by the Infomap yields
description length of 5.7 bits per step
(see codelength in Listing 3).
While these approaches are
relatively efficient, they are not
appropriate in all situations.
Recent algorithms aim to identify
hierarchically nestedas well as
overlappingcommunities, where an
individual can be placed in more than
one community or no community at
all. There is also a great emphasis on
efficiency of community detection
tools as large networks are becoming
increasingly common in many areas.
References
[1] Barabsi, A. L., Gulbahce, N., & Loscalzo, J. Network
medicine: a network-based approach to human
disease. Nature Reviews Genetics, 12, 1 (2011),
5668.
[2] Goh, K. I., Cusick, M. E., Valle, D., Childs, B., Vidal,
M., & Barabsi, A. L. The human disease network.
Proceedings of the National Academy of Sciences,
104, 21) (2007), 86858690.
[3] Rosvall, M., & Bergstrom, C. T. Maps of random
walks on complex networks reveal community
structure. Proceedings of the National Academy of
Sciences, 105, 4 (2008), 11181123.
[4] Raghavan, U. N., Albert, R., & Kumara, S. Near linear
time algorithm to detect community structures
in large-scale networks. Physical Review E, 76, 3
(2007), 036106.
[5] Pons, P., & Latapy, M. Computing communities in
large networks using random walks. Journal of
Graph Algorithms and Applications, 10, 2 (2006),
191218.
Copyright held by Owner(s)/Author(s).
POINTERS
HEALTH 2.0
Health informatics are a big part
of the wearable tech boom, but
are increasingly pervasive in the
tech landscape. According to
Wikipedia, [Health informatics]
deals with the resources, devices,
and methods required to optimize
the acquisition, storage, retrieval,
and use of information in health and
biomedicine. This is particularly
important as machine learning and
data analysis algorithms are able to
produce remarkable insights. The
future is here, with watches and body
monitors hoping to predict a heart
attack before it happens. Here are a
few resources to get you started.
Ashok Rao
HEALTH TECH STARTUPS

Predilytics
Part of dealing with information is
knowing how to use it well. Sometimes
thats difficult given the scattered
nature of various feeds, without any
common thread. Predilytics aims
to use data to answer healthcares
most important questions. Applying
preditictive analytics to healthcare
may prove promising not only for
Predilyticss bottomline, but for the
healthcare industry as a whole.
http://www.predilytics.com
Disruptive Healthcare Start-up
Oscar Raises $80 Million,
Valuation Nears $1 Billion
By Steven Bertoni
Maybe HealthCare.gov is the clearest
example of the lack of quality
technology in health insurance. In
fact, obtaining information from
pretty much any insurer is painfully
difficult, bureaucratic, and the
transaction feels like its from the
1950s. Oscar aims to change this by
applying simple, mobile-integrated,
ACRONYMS
technology to make access to

healthcare information timely and
easy.
http://www.forbes.com/sites/
stevenbertoni/2014/05/14/disruptivehealthcare-start-up-oscar-raises-80million-valuation-nears-1-billion/.
IN MEMORIAM
Maybe the person who knows best
about Health 2.0 is someone who
founded and ran the countrys largest
provider of managed, medical care.
Morris F. Collen, who co-founded
Kaiser Permanente in the 1940s, was a
champion of computerized medicine.,
He designed the first system for
automating multiphasic health
checkups. Dr. Collen passed away on
September 27, 2014 and is a hero to
many in this field.
http://www.nytimes.com/2014/10/05/us/
morris-collen-computerized-medicinepioneer-dies-at-100.html?_r=0
HEALTH INFORMATICS ABROAD

Asia Pacific Association
for Medical Informatics
While improvements in technology
are going a long way to systematized
healthcare in the U.S., a shocking
dearth of basic, computerized
processes abroad is a critical
challenge for many poorer countries.
Rapid advances in mobile
technology are particularly helpful
in poorer countries where remote
villages, without access to modern
technology, can obtain and transmit
health information (including
critical prescription data) quickly and
efficiently, saving lives in the process.
APAMI was formed in 1993 as
the Asia and Pacific regional branch
of IMIA (International Medical
Informatics Association). While the
APAMI 2014 Conference has past, be
sure to check back next year.
http://www.apami.org
AMIA American Medical Informatics

Association: A non-profit organization,
founded in 1989 as a result of
the merger of other NGOs. AMIA
supports the development and
application of health and biomedical
informatics for the improvement of
research, teaching, patient care, and
administration in healthcare.
CDSS Clinical Decision Support
System: It is an expert system that
links health data with appropriate
health observations, which in turn
supports doctors, clinicians, and
patients with medical decision-making.
CDW Clinical Data Warehouse (also
Clinical Data Repository): It is a real-time
database that integrates information
from multiple sources to present a
unified, clear medical history of (usually)
a single patient. Typical sources of
data for the CDW include medical test
reports, hospital admission, progress
notes, discharge details, etc.
HITECH Act Health Information
Technology for Economics and Clinical
Health Act, USA: Enacted as a part of the
American Recovery and Reinvestment
Act of 2009, this act aims to promote
the adoption and meaningful use of
health information technology. It also
addresses various concerns regarding
privacy and security.
MEDLINE Medical Literature Analysis
and Retrieval System Online (also
MEDLARS Online): It is a freely available,
online database, maintained by the U.S.
National Library of Medicine. It contains
life sciences and biomedicine related
data in the form of journal citations and
abstracts from academic and scientific
literature around the world.
TBI Translational Bioinformatics:
One of the sub-fields of health
informatics, it is a combination of
clinical informatics, biostatistics, and
molecular bioinformatics. It supports
the storage, analysis, and integration
of massive amounts of biomedical and
genomic data using data miningand
other data-based technologiesto
formulate disease diagnosis strategies,
health outcome predictions, treatment
therapies, etc.
61
end
EVENTS
CONFERENCES
International Conference on
Distributed Computing and
Networking
Birla Institute of Technology &
Science, Pilani - K K Birla Goa Campus
Goa, India
January 4-7, 2015
http://www.icdcn.org
Pacific Symposium on Biocomputing
Fairmont Orchid
Kohala Coast, HI
January 4-8, 2015
http://psb.stanford.edu
Keystone Symposia: Precision Genome
Engineering and Synthetic Biology
Big Sky Resort
Big Sky, MT
January 11-16, 2015
http://www.keystonesymposia.org
International Joint Conference
on Biomedical Engineering Systems
and Technologies
Sana Lisboa Hotel
Lisbon, Portugal
January 12-15, 2015
http://www.biostec.org
International Conference on Tangible,
Embedded, and Embodied Interaction
Stanford University
Stanford, CA
January 15-19, 2015
http://www.tei-conf.org/15
International Conference on
High-Performance and Embedded
Architectures and Compilers
Forum Centre at Amsterdam RAI
Amsterdam, Holland
January 19-21, 2015
http://www.hipeac.net/2015/amsterdam
62
SIGAI Career Network and Conference

Radisson Hotel & Suites Austin
Downtown
Austin, TX
January 26, 2015
http://sigai.acm.org/cnc
Eighth ACM International Conference
on Web Search and Data Mining
Crowne Plaza Shanghai Fudan
Shanghai, China
February 2-6. 2015
http://www.wsdm-conference.org/2015
International Conference
on Information Systems Security
and Privacy
ESEO Grande Ecole dingnieurs
Angers, France
February 9-11, 2015
http://www.icissp.org
Sixteenth Workshop on Mobile
Computing Systems and Applications
Santa Fe, NM
February 12-13, 2015
http://www.hotmobile.org/2015
Third ACM/SIGDA International
Symposium on Field-Programmable
Gate Arrays
Monterey Conference Center
Monterey, CA
http://www.eecs.ucf.edu/isfpga
Second International Conference on
Perception and Machine Intelligence
Saha Institute of Nuclear Physics
Kolkata, India
http://www.permin.in/permin15
46th ACM Technical Symposium on
Computer Science Education
Kansas City Convention Center
Kansas City, MO
March 4-7, 2015
http://sigcse2015.sigcse.org
Seventh International Conference on

Bioinformatics and Computational
Biology
Waikiki Beach Marriott Resort & Spa
Honolulu, HI
March 9-11, 2015
http://www.cs.umb.edu/bicob
Sixth International Conference
of Augmented Human
Marina Bay Sands Expo and
Convention Centre
Singapore, Singapore
March 9-11, 2015
http://asg.sutd.edu.sg/ah2015/home
Architectural Support
for Programming Languages
and Operating Systems
Istanbul, Turkey
March 14-18, 2015
http://asplos15.bilkent.edu.tr
14th International Conference
on Modularity
Colorado State University
Fort Collins, CO
March 16-19, 2015
http://www.aosd.net/2015
CONTESTS & EVENTS

Facebook Hacker Cup
The Facebook Hacker Cup is a
worldwide computer programming
competition that takes place each
year. The contest features algorithmic
problems and several rounds of
challenges in order to take home the
coveted Hacker Cup Trophy as well as
cash prizes. The qualification period
begins in early January 2015 culminating
with the final round at Facebooks
headquarters in Menlo Park, CA.
https://www.facebook.com/hackercup/
Launch48 Weekend Accelerators
Throughout November 2014 to
February 2015, Oxygen Startups will
be featuring Launch48 Weekends
FEATURED EVENT
Accelerators. In this program,

participants pitch, build, and launch
a startup within 48 hours. Members
will work with mentors and peers
to develop business and customer
plans. Other Oxygen Startups
programs include accelerators to
help existing, but early, startups find
guidance and funding.
http://oxygenstartups.com/
CS Games
CS Games is an annual computer
science competition that will take
place next year at Sherbrooke
University in Canada from March 13
to 16. The competition is open to any
university and each university can
register teams of six to 10 members.
Within the competition, there are
events focusing on various aspects of
computer science such as artificial
intelligence, algorithms, gaming,
scavenger hunts, and more. Trophies
and awards will be presented at the
closing ceremony.
http://csgames.org/
jQuery UK 2015
On March 6, 2015, a conference will be
taking place in the UK for the jQuery
framework. jQuery is a very popular
frontend JavaScript framework used
for interactive websites, animations,
and graphical interfaces. The event
features several industry speakers
from well-known companies such as
Google, Twitter, and GitHub.
http://jqueryuk.com/2015/
Photo by Leung Cho Pan
GRANTS AND SCHOLARSHIPS

AACE International Scholarship
Website: http://www.aacei.org/awards/
scholarships/overview.shtml
Deadline: Feburary 7, 2015
Eligibility: Full-time students in a
field related to cost-engineering or
cost-management, such as electrical
engineering or computer science, with
a GPA of 3.0 or better

Benefits: $2,000 - $8,000
Explanation: In line with its mission to
advance the study of cost-engineering
and cost-management through the
total cost management process,
AACE International awards hardworking students in related programs.
Spencer Foundation
Small Research Grant
Website: http://www.spencer.org/
content.cfm/budgets-50000-or-less
Deadline: Multiple Deadlines
throughout 2015; February 5th, June
2nd, and August 20th.
Eligibility: Grants are awarded for
research related to any of Spencers
areas of inquiry: education and
social opportunity; organizational
learning; purposes and values of
education; and teaching, learning,
and instructional resources.
Benefits: Project budgets up to $50,000
Explanation: Established by Lyle M.
Spencer, the Spencer Foundation aims
to investigate ways education can be
improved around the world.
Department of Energy Computational
Science Graduate Fellowship
Website: https://www.krellinst.org/csgf/
how-apply
Deadline: January 2015
Eligibility: U.S .citizens and
permanent residents planning study
toward a Ph.D at an accredited U.S.
university, and have not yet started
their second year of doctoral studies.
Benefits: $36,000 stipend, tuition, and
fees for up to four years. An allowance
for a computer workstation is added.
Explanation: The Computations
Science Graduate Fellowship is meant
to encourage interdisciplinary work
and collaboration in the field of
computational science. The fellowship
includes a practicum at a Department
of Energy laboratory. A conference is
held each summer for recipients.
Third International Conference on

Bioinformatics and Computational
Biology (ICBCB 2015)
Hong Kong, China
March 12-13, 2015
Bioinformatics uses principles
of computer science and
mathematics to interpret and
study data in biological research.
This has allowed researchers and
practitioners to make tremendous
strides, especially in the fields
of genetics and genomics. As
research problems continue to
become harder and more complex,
developments in computing
will be critical to the future of
biological research.
In March, the South Asia
Institute of Science and
Engineering (SAISE) and WIT
Press will host ICBCB 2015 in
Hong Kong. The conference
welcomes the latest research
in bioinformatics and bridges
the gap between academia and
industry. Topics of interest
will include sequence analysis,
biological data analysis, and
genetics. Attendees of the
conference will be able to step
into Hong Kong, which offers
a bustling environment and
a beautiful skyline.
Learn more at http://www.icbcb.
org/.
Rohit
Goyal
63
end
BEMUSEMENT
From A to B and Back Again
Puzzles:
Thomas
the Truant
Thomas has missed an excessive
number of days of school, so he
must meet with Principal Davis.
Mr. Davis asks him, Why on Earth
have you missed so many days?
Thomas replies: There just
isnt enough time for school. I
need 8 hours of sleep a day, which
adds up to about 122 days a year.
Weekends off is 104 days a year.
Summer vacation is about 60 days.
If I spend about an hour on each
meal, thats 3 hours a day or 45
days a year. I need at least 2 hours
of exercise and relaxation time
each day to stay physically and
mentally fit, adding another 30
days. Add all of that up and you get
about 361 days. That only leaves 4
days for school.
The principal knows Thomas is
full of it, but cant figure out why.
Why is Thomas wrong?
Source: http://goodriddlesnow.com/riddles/
view/743
Post-Bachelors Disorder
Find the solution at: http://xrds.acm.

org/bemusement/2014.cfm
PhD Comics Jorge Cham
SUBMIT A PUZZLE
64
Can you do better?

Bemusements would like your
puzzles and mathematical games
(but not Sudoku). Contact
xrds@acm.org to submit yours!
acm
STUDENT MEMBERSHIP APPLICATION

CODE: CRSRDS
Name
Join ACM online: www.acm.org/joinacm
Please print clearly
INSTRUCTIONS
Address
City
State/Province
Country
E-mail address
Area code & Daytime phone
Mobile phone
Carefully complete this application and return

with payment by mail or fax to ACM. You must
be a full-time student to qualify for student rates.
Postal code/Zip
CONTACT ACM
Member number, if applicable
MEMBERSHIP BENEFITS AND OPTIONS

Free software and courseware through the ACM
Academic Initiative
Free e-mentoring services from MentorNet
Electronic subscriptions to Communications of the ACM
and XRDS: Crossroads magazines
Online courses, online books and videos
ACM's CareerNews (twice monthly)
ACM e-news digest TechNews (thrice weekly)

ACM online newsletter MemberNet (monthly)
Student Quick Takes, ACM student e-newsletter (quarterly)
Free "acm.org" email forwarding address plus filtering
through Postini
Option to subscribe to the full ACM Digital Library
Discounts on ACM publications and conferences,
valuable products and services, and more
PLEASE CHOOSE ONE:

Student Membership: $19 (USD)
Student Membership PLUS Digital Library: $42 (USD)
Student Membership PLUS Print CACM Magazine: $42 (USD)
Student Membership w/Digital Library PLUS Print CACM Magazine: $62 (USD)
P U B L I C AT I O N S
Check the appropriate box and calculate
amount due on reverse.
ACM Inroads
Communications of the ACM
Computers in Entertainment (online only)
Computing Reviews
Computing Surveys
Evolutionary Computation (MIT Press)
interactions, new visions of human-computer interaction
(included in SIGCHI membership)
Intl Journal of Network Management (online only) (Wiley)
Intl Journal on Very Large Databases
Journal of Educational Resources in Computing (see TOCE)
Journal of Experimental Algorithmics (online only)
Journal of Personal and Ubiquitous Computing
Journal of the ACM
Journal on Computing and Cultural Heritage
Journal on Data and Information Quality
Journal on Emerging Technologies in Computing Systems
Linux Journal (SSC)
Mobile Networks and Applications
Wireless Networks
XRDS (included with membership)
Transactions on:
Accessible Computing
Algorithms
Applied Perception
Architecture & Code Optimization
Asian Language Information Processing
Autonomous and Adaptive Systems
Computational Biology and Bioinformatics
Computer-Human Interaction
Computational Logic
Computation Theory
Computer Systems
Computing Education (formerly JERIC)
Database Systems
Design Automation of Electronic Systems
Economics and Computation
Embedded Computing Systems
Graphics
Information and System Security
Information Systems
Intelligent Systems and Technology
Interactive Intelligent Systems
Internet Technology
Knowledge Discovery From Data
Management Information Systems
Mathematical Software
Modeling and Computer Simulation
Multimedia Computing, Communications, and Applications
Networking
Programming Languages & Systems
Reconfigurable Technology & Systems
Sensor Networks
Software Engineering and Methodology
Speech and Language Processing (online only)
Storage
Web
Marked are available in the ACM Digital Library
* Check here to have publications delivered via Expedited Air Service.
For residents outside North America only.
Please check
Issues
per year
4
12
4
12
4
4
6
Code
178
101
247
104
103
177
123
Member Rate
$16
$25
$43
$55
$36
$32
$22
Air Rate*
$58
$58
N/A
$39
$32
$30
$35
6
4
N/A
12
6
6
4
4
4
12
6
4
4
136
148
N/A
129
144
102
173
171
154
137
130
125
XRoads
$92
$83
N/A
$31
$65
$55
$49
$49
$42
$31
$72
$72
$39
$30
$30
N/A
N/A
$30
$58
$23
$25
$23
$33
$29
$29
N/A
4
4
4
4
4
4
4
4
4
8
4
174
151
145
146
138
158
149
119
135
176
114
277
109
128
192
142
112
134
113
179
191
140
170
190
108
116
156
118
110
172
155
115
253
157
159
$24
$23
$23
$23
$23
$23
$49
$25
$25
$32
$25
N/A
$25
$25
$23
$23
$25
$23
$25
$72
$76
$23
$23
$22
$25
$25
$23
$52
$32
$24
$23
$25
N/A
$23
$23
phone: 800-342-6626
(US & Canada)
+1-212-626-0500
(Global)
hours: 8:30am4:30pm
US Eastern Time
fax:
+1-212-944-1318
email: acmhelp@acm.org
mail:
Association for Computing
Machinery, Inc.
General Post Office
P.O. Box 30777
New York, NY 10087-0777
For immediate processing, FAX this
application to +1-212-944-1318.
PAYMENT INFORMATION
Payment must accompany application
$
Member dues ($19, $42, or $62)
To have Communications of the ACM
sent to you via Expedited Air Service,
add $58 here (for residents outside of
$
North America only).
Publications
Total amount due
Check or money order (make payable to ACM,

Inc. in U.S. dollars or equivalent in foreign currency)
Visa/Mastercard
American Express
Card number
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
6
6
4
4
4
4
4
4
$49
$52
$43
$43
$39
$41
$20
$43
$44
$49
$47
$25
$46
$43
$49
$44
$51
$44
$47
$46
$49
$42
$50
$47
$47
$51
$42
$29
$59
$49
$42
$43
$33
$42
$41
Signature
Member dues, subscriptions, and optional contributions
are tax deductible under certain circumstances. Please
consult with your tax advisor.
EDUCATION
Name of School
Please check one: High School (Pre-college, Secondary
School) College: Freshman/1st yr. Sophomore/2nd yr.
Junior/3rd yr. Senior/4th yr. Graduate Student:
Masters Program Doctorate Program Postdoctoral
Program Non-Traditional Student
Major
Expected mo./yr. of grad.
Age Range: 17 & under 18-21 22-25 26-30

31-35 36-40 41-45 46-50 51-55 56-59 60+
Do you belong to an ACM Student Chapter? Yes No
I attest that the information given is correct and that I will
abide by the ACM Code of Ethics. I understand that my
membership is non transferable.
Signature
PUBLICATION SUBTOTAL:
Exp. date
CAREERS at the NATIONAL SECURITY AGENCY
EXTRAORDINARY WORK
Inside our walls, you will find the most extraordinary people doing
the most extraordinary work. Yet its not just finite field theory,
quantum computing or RF engineering. Its not just discrete
mathematics or graph analytics.
Its all of these and more rolled up into an organization that leads
the world in signals intelligence and information assurance.
Inside our walls you will find extraordinary people, doing
extraordinary work, for an extraordinary cause:
The safety and security of the United States of America.
U.S. citizenship is required. NSA is an Equal Opportunity Employer. All applicants for employment are considered without
regard to race, color, religion, sex, national origin, age, marital status, disability, sexual orientation, or status as a parent.
Computer/Electrical Engineering
Computer Science
Cybersecurity
Information Assurance
Mathematics
Foreign Language
Intelligence Analysis
Cryptanalysis
Signals Analysis
Business Management
Finance & Accounting
Paid Internships, Scholarships & Co-op
APPLY TODAY
Search NSA to Download
WHERE INTELLIGENCE GOES TO WORK

Crossroadsxrds2014winter DL

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Crossroadsxrds2014winter DL

Diunggah oleh

Hak Cipta:

Format Tersedia

XRDS

Crossroads The ACM Magazine for Students

Are you looking for

A highly targeted focus on job

Job Alert system that notifies you of

Access to hundreds of industry job postings

Resume posting keeping you connected

Career coaching and guidance available

Free access to a content library of the best

Career & Job Center at:

Visit today at http://jobs.acm.org

Who can participate?

New and recent Ph.Ds, doctoral candidates, other graduate students

What is the schedule?

Application deadlineFebruary 28, 2015.

PHOTOS: HLFF/B. Kreutzer (2)

My $300 Home Cloud Server:

Image by Valentina Photo / Shutterstock.com

An Antidote to Impostor Syndrome

Gathering People to Gather Data

Opportunities of Social Media

Did I Take My Meds Today?

The Regenstrief Global Health

Here Comes the #Engagement:

Wearable Technologies: One step

Challenges in Personal Health

Trevor van Mierlo: The Story

Your future career, secured.

Full federal scholarships available for U.S. citizens

Laurie Faith Cranor,

Erin Claire Carson

Noam Nisan, Hebrew

Students and faculty can take advantage

Panagiotis Takis Metaxas ,

Issue Feature Editor

Distinguished Speakers Program

XRDS Managing Editor &

PUBLIC AT IONS BOA RD

LETTER FROM THE EDITORS

of research that was funded by the

fessor Robert Full, a biologist at UC

This formative basic research may

about the Large Hadron Collider at

social networks and understanding

ON THE NLP ISSUE

NEW ON THE BLOG

that its pulling an absurd

How to contact XRDS: Send a letter to the

In the U.S., although medical records contain patient

ts the crack of dawn,

(e.g., the increasing affordability and sophistication

The idea that

the form of plain text, survey responses, or even raw

helped doctors make better

Photo by Katherine Breeden

A data silo refers to records

Image by Slavoljub Pantelic

ho wants to set breakpoints when they can

cessor performance and the trend is

simple scripts to the mix and before

which are advertised, carried out,

n this edition of XRDS, I would like

The Cyborg Foundation is founded by Neil Harbisson, who is known as the