In this talk, I present an optimal matching framework for Hybrid Clinical Trials (HCTs),which combine randomized controlled trials (RCT) with real-world electronic health record (EHR) data to estimate population treatment effects using hybrid control arms .HCTs can improve trial efficiency, generalizability, and precision, while reducing costs. We formulate hybrid cohort construction as a bipartite matching problem and solve it using linear programming. A weighted objective function jointly reduces internal bias between the treatment and hybrid control arms while promoting generalizability to the target population. We evaluate the approach using FRESCA simulations parameterized with data from the ALLHAT and SPRINT RCTs, and extend FRESCA to quantify internal validity via cohort internal disparity. Across simulation studies, HCT optimal-matching variants yield accurate population hazard ratio estimates and improved internal validity compared with baseline matching approaches. These findings suggest that optimal matching provides a flexible and promising methodology for estimating population treatment effects when integrating RCT and EHR data.
About the Speaker 
Corey is a fourth year PhD student in the Department of Mathematical Sciences at Rensselaer Polytechnic Institute. His research focuses on applying optimization techniques to public health problems.
