Biostatistics for Dummies

Biomedical Computing Cross-Training Seminar October 18th, 2002

What is Biostatistics?
Techniques
Mathematics l Statistics l Computing
l

Data
Medicine l Biology
l

What is Biostatistics?
Knowledge of biological process

Biological data

Common Applications
(Medical and otherwise) Clinical medicine Epidemiologic studies Biological laboratory research Biological field research Genetics Environmental health Health services Ecology Fisheries Wildlife biology Agriculture Forestry

Biostatisticians Work
Develop study design Conduct analysis Oversee and regulate Determine policy Training researchers Development of new methods

Some Statistics on Biostatistics

Internet search (Google) > 210,000 hits > 50 Graduate Programs in U.S.

Too much to cover in one hour!

Center Focus
MSU strengths
l

Computational simulation in physical sciences Environmental health sciences

Computational simulation in environmental health sciences

l

Bioinformatics is crowded

Build on appreciable MSU strength Establish ourselves

l l

Unique capability Particular appeal to NIEHS

Focus of Seminar
Statistical methodologies
Computational simulation in environmental health sciences l Can be classified as biostatistics
l

Stochastic modeling
Time series l Spatial statistics*
l

The Application
Of interest
l l

Objectives
l

Cancer incidence rate Pesticide exposure Age Gender Race Socioeconomic status

Of concern
l l l l

Suitably adjust cancer incidence rate Determine if relationship exists Develop model
l l l

Explain relationship Estimate cancer rate Predict cancer rate

The Data
MS State Dept. Health Central Cancer Registry (1996 1998, by person)
l

N.S.S. & U.S. Dept. of Commerce National T.I.S. (1972-2001, by county)

l

l l l l l

Tumor type Age Gender Race County of residence Cancer morbidity

l

Number of acres harvested Type of crop

Crude incidence/100,000 Age adjusted incidence/100,000

Why (Bio)statistics?
Statistics
l l

Entropy

Science of uncertainty Model order from disorder Large scale rational explanation Smaller scale residual uncertainty

Disorder exists
l

Chaos

x0 Deterministic equation Randomness

(Bio)statistical Data
Independent identically distributed Inhomogeneous data Dependent data
Time series l Spatial statistics
l

Time Series
Identically distributed Time dependent Equally spaced
Randomness

Objectives in Time Series

Graphical description
Time plots l Correlation plots l Spectral plots
l

Modeling Inference Prediction

Time Series Models

Linear Models Covariance stationary
l l

e(t) ~ i.i.d
l l

Constant mean Constant variance Covariance function of distance in time

Zero mean Finite variance

f square summable

Nonlinear Time Series

Amplitude-frequency dependence Jump phenomenon Harmonics Synchronization Limit cycles Biomedical applications
l l l

l l

Respiration Lupus-erythematosis Urinary introgen excretion Neural science Human pupillary system

Some Nonlinear Models

Nonlinear AR
l

Additive noise AR Smoothed TAR Markov chain driven Fractals

Threshold
l l l l

Amplitudedependent exponential AR Bilinear AR with conditional heteroscedasticity Functional coefficient AR

A Threshold Model

A Threshold Model

Describing Correlation
Autocorrelation
AR: exponential decay l MA: 0 past q
l

Partial autocorrelation
AR: 0 past p l MA: exponential decay
l

Cross-correlation Relationship to spectral density

Spatial Statistics*
Data components
l

Data structures
l l l

Spatial locations S = {s1,s2,,sn} Observable variable {Z(s1),Z(s2),,Z(sn)} s D Rk

Correlation

Geostatistical Lattice Point patterns or marked spatial point processes Objects

Assumptions on Z and D

Biological Applications
Geostatistics
l l

Soil science Public health Remote sensing Medical imaging Tumor growth rate In vitro cell growth

Lattice
l l

Point patterns
l l

Spatial Temporal Models

Combine time series with spatial data Application
l

Time element
l

Pesticide exposure

time

develop cancer

Spatial element
l

Proximity to pesticide use