A1082
Title: Heterogeneous causal mediation analysis using Bayesian additive regression trees
Authors: Xu Qin - University of Pittsburgh (United States) [presenting]
Jiebiao Wang - University of Pittsburgh (United States)
Chen Liu - University of Pittsburgh (United States)
Abstract: Causal mediation analysis provides insights into the mechanisms through which treatments affect outcomes. While mediation effects often vary across individuals, most existing methods focus solely on population-average effects, overlooking individual-level heterogeneity. To address this limitation, a Bayesian regression tree ensemble method is proposed that flexibly models non-linear relationships and captures treatment-by-mediator interactions in the mediation process. Using hierarchical posterior sampling, the approach provides credible intervals with nominal coverage rates for testing heterogeneous mediation effects. Additionally, regression tree summaries are leveraged to identify subgroups with distinct mediation effects, and SHapley Additive exPlanation (SHAP) values are employed to highlight key moderators and their influence on the mediation process. Comprehensive simulations demonstrate the method's accuracy in estimating and inferring heterogeneous mediation effects. Finally, the method is applied to investigate the heterogeneous mediation of Alzheimer's disease pathology burden on the effect of apolipoprotein E (APOE) genotype on late-life cognition.
