Bayesian Clustering and Sparse Modeling of Interactions in Epidemiology

• Activity Number: 224
• Type: Topic Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Amy H. Herring*+ and David Dunson
• Companies: The University of North Carolina at Chapel Hill and Duke University
• Keywords: Bayesian clustering ; latent class models ; reproductive epidemiology ; dimensionality reduction

Abstract

Often we wish to study the relationship between a high-dimensional set of related exposures and a high-dimensional set of related health outcomes. For example, many different genetic and behavioral factors may contribute to risk different birth defects. Analyses that assess the relationship of a single outcome and a few exposures and confounders in isolation can be misleading. To build accurate predictive models and better understand the etiology of complex health outcomes, it is crucial to consider many different exposures, confounders, interactions, and outcomes together in a multivariate analysis; however, dimensionality reduction methods are typically needed. We develop and apply novel Bayesian latent class methods for dimensionality reduction motivated specifically by the challenges arising in birth defects studies.