|
Abstract:
|
Comparative effectiveness research (CER) is designed to synthesize evidence of the benefits and harms of a treatment option from disparate sources including randomized clinical trials, observational studies and registry studies. The task of addressing study-specific heterogeneities is one of the most difficult challenges in CER. Bayesian hierarchical model with non-parametric extension provide a powerful and convenient platform that formalizes the information borrowing strength across the studies. In this paper, we propose a propensity score-based Bayesian non-parametric Dirichlet process mixture model that summarizes information from multiple observational studies to draw inference on the causal treatment effect. Simulation studies are conducted to evaluate the model performance under different scenarios.
|
