B0978
Title: On heterogeneous treatment effects in heterogeneous causal graphs
Authors: Hengrui Cai - University of California Irvine (United States) [presenting]
Abstract: Heterogeneity and comorbidity are two interwoven challenges associated with various healthcare problems that greatly hampered research on developing effective treatment and understanding of the underlying neurobiological mechanism. Very few studies have been conducted to investigate heterogeneous causal effects (HCEs) in graphical contexts due to the lack of statistical methods. To characterize this heterogeneity, heterogeneous causal graphs (HCGs) are first conceptualized by generalizing the causal graphical model with confounder-based interactions and multiple mediators. Such confounders with an interaction with the treatment are known as moderators. This allows for flexible production of HCGs given different moderators and explicitly characterize HCEs from the treatment or potential mediators on the outcome. The theoretical forms of HCEs are established and their properties are derived at the individual level in both linear and nonlinear models. An interactive structural learning is developed to estimate the complex HCGs and HCEs with confidence intervals provided. The method is empirically justified by extensive simulations and its practical usefulness is illustrated by exploring causality among psychiatric disorders for trauma survivors.