A0327
Title: A flexible Bayesian g-formula for causal survival analyses with time-dependent confounding
Authors: Fan Li - Yale University (United States) [presenting]
Xinyuan Chen - Mississippi State University (United States)
Liangyuan Hu - Rutgers University (United States)
Abstract: In longitudinal observational studies with a time-to-event outcome, a common objective in causal analysis is to estimate the causal survival curve under hypothetical intervention scenarios within the study cohort. The g-formula is a particularly useful tool for this analysis. To enhance the traditional parametric g-formula approach, a more adaptable Bayesian g-formula estimator is developed. It incorporates Bayesian additive regression trees in the modeling of the time-evolving generative components, aiming to mitigate bias due to model misspecification. Specifically, a more general class of g-formulas is introduced for discrete survival data. These formulas can incorporate the longitudinal balancing scores, which serve as an effective method for dimension reduction and are vital when dealing with an expanding array of time-varying confounders. The minimum sufficient formulation of these longitudinal balancing scores is linked to the nature of treatment regimes, whether static or dynamic. For each type of treatment regime, posterior sampling algorithms are provided, which are grounded in the Bayesian additive regression trees framework. Simulation studies are conducted to illustrate the empirical performance of the proposed Bayesian g-formula estimators and compare them with existing parametric estimators. The practical utility of the methods is further demonstrated in real-world scenarios using data from the Yale New Haven Health System's electronic health records.
