Abstract:
Causal mediation analysis decomposes a treatment's effect into components operating through distinct causal pathways and characterizes the mediator--outcome relationship. In time-to-event settings, competing events, which preclude the primary endpoint, substantially complicate this decomposition. We provide a unified treatment of causal mediation analysis for a time-to-event endpoint under competing events. We show that the total effect on the cause-specific cumulative incidence decomposes into a direct effect, an indirect effect through the mediator, and an indirect effect through the competing events. We develop this decomposition under the natural, stochastic interventional, and separable effects frameworks, and show that in certain settings all three yield identical statistical functionals despite distinct identification assumptions. We derive efficient influence functions for the counterfactual cumulative incidence curves and the decomposed effects, and propose multiply robust double-machine-learning estimators that attain the semiparametric efficiency bound, together with a sensitivity analysis for unmeasured common causes of the time-to-event endpoints. We further study the mediator--outcome relationship via kernel-smoothed estimands. We apply the methods to immune correlates analyses of COVID-19 vaccine and HIV-1 prevention trials.
About the Speaker:
Dr. Yuhao Deng is currently a postdoctoral research fellow at Fred Hutchinson Cancer Center. He received his PhD at Peking University and previously held a postdoctoral fellowship at the University of Michigan before his current position. His research spans include causal inference, survival analysis, semiparametric modeling, and econometrics, with a particular focus on causal mediation methods in the presence of unmeasured confounding and intercurrent events. He has published more than 20 papers in journals including The Annals of Applied Statistics, Biometrics, Journal of Statistical Software, and Communications Biology. He serves as a reviewer for over 20 journals. He has been granted four patents and co-translated three books on statistics.

Your participation is warmly welcomed!

欢迎扫码关注北大统计科学中心公众号,了解更多讲座信息!