A0168
Title: High dimensional mediation analysis for longitudinal mediators and survival outcomes
Authors: Lei Liu - Washington University in St. Louis (United States) [presenting]
Abstract: Mediation analysis with high-dimensional mediators is important in identifying epigenetic pathways between environmental exposures and health outcomes. However, high-dimensional mediation analysis methods for longitudinal mediators and survival outcomes remain underdeveloped. This gap is addressed by proposing a methodology that accommodates time-varying mediators, characterized through multivariate longitudinal observable variables, to explore mediation effects over time. The approach uses a longitudinal mixed effects model to examine the relationship between exposure and the mediating process. The mediating process is connected to survival outcomes using a Cox proportional hazards model with time-varying mediators. To manage high-dimensional data, a mediation-based sure independence screening method is first employed for dimension reduction. A Lasso inference procedure is utilized to identify the significant time-varying and heterogeneous mediators. A multiple-testing procedure is adopted to accurately control the family-wise error rate when testing high-dimensional mediation hypotheses. Simulation studies and analysis of the Coronary Artery Risk Development in Young Adults (CARDIA) Study demonstrate the utility and validity of the method.
