Bayesian Variable Selection Methods for Matched Case-Control Studies

• Activity Number: 628
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Josephine Asafu-Adjei*+ and Mahlet Tadesse and Brent A. Coull and Raji Balasubramanian and Rebecca A. Betensky
• Companies: Harvard School of Public Health and Georgetown University and Harvard School of Public Health and Division of Biostatistics and Epidemiology and Harvard School of Public Health
• Keywords: Bayesian analysis ; Conditional logistic regression ; Matched case-control studies ; Variable selection methods

Abstract

Matched case-control designs are currently used in many biomedical applications. To ensure high efficiency and statistical power in identifying features that best discriminate cases from controls, it is important to account for the use of matched designs. However, in many instances, particularly in high dimensional cases, few variable selection methods account for matching. Bayesian approaches to variable selection are beneficial in that they offer efficient methods for handling high dimensional biological data. They yield tractable models that incorporate the biological structure of the data through the selection of prior distributions. We propose a novel variable selection method to account for matching in case-control studies by formulating conditional logistic regression in a Bayesian context. This methodology is carefully developed to handle various settings encountered in biomedical studies, including high dimensional data settings. Through simulation studies and matched studies examining different biomarkers, our proposed approach can be shown to yield improved results compared with more traditional variable selection methods.