Mediation analysis with high dimensional mediators and high dimensional exposures for genetic applications

Mediation analysis has been a popular framework for elucidating the mediating mechanism of the exposure effect on the outcome in many disciplines including genetic studies. In genetic applications, both of the mediator and exposure may be high dimensional. Previous literature has focused on the setting with high dimensional mediator and low-dimensional exposure. In this talk, I present my works on two mediation problems with high dimensional exposure and mediators. The first paper has developed MedFix and MedMix, two procedures for mediator selection. The second paper has proposed MedDiC for the estimation and inference of the mediation effect from each exposure. In both papers, real data applications have yielded reproducible and biologically meaningful results.

Bio
Qi Zhang is an associate professor in statistics in the Department of Mathematics and Statistics at University of New Hampshire (UNH). He obtained his Ph.D. at the Department of Mathematics and Statistics, University of Pittsburgh. Zhang's research areas include high dimensional data analysis, causal inference, machine learning in general. His current research interests include high dimensional mediation, heterogeneous treatment effect estimation, high dimensional variable selection, and deep generative models. He is interested in statistical applications in multi-omics, medical records, and materials research.