Bayesian Predictive Outcome-Modeling Approach to Identifying Treatment Effect Heterogeneity for Continuous Outcomes in Clinical Trials

Abstract

Evaluating treatment efficacy in clinical trials relies on the average treatment effect (ATE) to inform clinical decision-making, but this may obscure meaningful variability in individual responses, often defined as treatment effect heterogeneity (TEH). Accurately evaluating this variability is useful for accurately assessing the impact of interventions and promoting personalized patient recommendations. Current methodologies may suffer from low power to detect TEH, inflated type I error, and limited interpretability.