Title: Bayesian feature selection in joint models with application to cardiovascular disease cohorts
Authors: Mirajul Islam - University of Florida (United States) [presenting]
Michael Daniels - University of Florida (United States)
Juned Siddique - Northwestern University (United States)
Abstract: Cardiovascular disease (CVD) cohorts collect data longitudinally to study the association between CVD event times and risk factors. An important area of scientific research is to understand better what features of CVD risk factor trajectories are associated with disease. We develop methods for feature selection in joint models where the features are viewed as a two-level variable selection problem with multiple features and multiple longitudinal processes. We modify a Bayesian sparse group selection prior for the joint modeling framework to select features both at the group level (CVD risk factor) and within a group (features of a longitudinal risk factor). We apply our methods to the Atherosclerosis Risk in Communities (ARIC) study data, a population-based and prospective cohort study consisting of 15,792 participants measured four times at three-year intervals, where it is important to investigate which characteristics of CVD risk factor trajectories are associated with death from CVD.