Predicting Veterans with Frequent Poor Mental Health

• Activity Number: 674
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Brady Stephens*+ and Robert Bossarte
• Companies: Department of Veterans Affairs and Department of Veterans Affairs
• Keywords: BRFSS ; classification ; Veterans ; mental health

Abstract

There are limited reliable measures of the number Veterans with frequent poor mental health within the U.S. population. The Centers for Disease Control and Prevention's (CDC) Behavior Risk Factor Surveillance System (BRFSS) is an annual source of self-reported Veteran status and their current health and well-being.

This report attempts to identify the most effective predictive classification technique for Veterans with frequent poor mental health based on demographics, health, and other important characteristics from the 2009 BRFSS data. Cross-validation will be used to confirm the selected model reduction procedure. Standard classification techniques - including principal components analysis, discriminant analysis, logistic regression, and support vector machine - will be considered. Training sets will be stratified by age and gender to account for the disparate composition between Veterans and Non-Veterans within the general population.

These findings provide a preliminary model for forecasting changes in the number of future Veterans with frequent poor mental health without regard for recent mental health initiatives among Veterans.