|
Abstract:
|
Reproducibility is a fundamental aspect of scientific research, but in healthcare research we have found numerous examples where the findings from randomized clinical trials (RCTs) are not able to be replicated in real-world data (RWD), or vice versa. One important contributing factor to this challenge is the unmeasured confounding, i.e., the variables available in RCTs are not available in RWD, and it could result in substantial bias in estimating treatment effect if not mitigated. In this presentation we will introduce a Bayesian hierarchical modeling method, along with simulation studies where external information is incorporated, to address this challenge.
|
