Lei Liu is Professor in the Division of Biostatistics at Washington University in St. Louis and associated editor of the revue Stat Med et Biometrics. The seminar will be held in the amphi Louis of ISPED at 3 p.m. on December 14, 2022.
Mediation analysis has been commonly used to study the effect of an exposure on an outcome through a mediator. In many studies, there exist high dimensional mediators, e.g., DNA methylations, microbiome, metabolites, and fMRI imaging. We are interested in high dimensional mediation analysis (HIMA) in these studies. We use a 3-step procedure to implement our methods: screening of marginal effects, estimation and inference for each mediator, and joint significance test for the FWER/FDR control. We have developed a variety of models to tackle challenging issues in HIMA, e.g.: different types of outcome (continuous, discrete, time to event), compositional features of the microbiome relative abundance data. We will use simulation and real data examples to illustrate our methods. R packages HIMA and HIMA2 has been developed to implement our methods.
The event can also be viewed on https://u-bordeaux-fr.zoom.us/j/81010001784