Trial Augmentation Using External Data and Models: Toward Harmony Between Observational Studies and Trials
Join us on Wednesday, February 4th for the Department of Epidemiology seminar series featuring Dr. Issa Dahabreh discussing Trial Augmentation Using External Data and Models: Toward Harmony Between Observational Studies and Trials
Abstract: We introduce trial augmentation, a new approach to analyzing randomized trials that leverages external data — either historical experimental data or observational data — to improve trial efficiency without sacrificing the unbiasedness guarantee provided by randomization. We characterize a broad class of randomization-aware estimators that integrate external data through data-adaptive models (e.g., machine learning or generative models), yielding higher efficiency and statistical power than estimators based on trial data alone. Crucially, members of this class exploit randomization to remain unbiased even when the external data are misaligned with the trial population or affected by unmeasured confounding. We show that several widely used estimators, including the efficient trial-only estimator, are special cases within this framework. We further demonstrate how combining two or more randomization-aware estimators yields procedures with two key properties: (1) robustness to misalignment and unmeasured confounding in the external data, and (2) efficiency that is at least as high as, and typically higher than, that of the component estimators. We situate these results within a broader research program aimed at a more harmonious integration of observational analyses and randomized trials.
Bio: Issa Dahabreh, MD ScD is Associate Professor of Epidemiology and Biostatistics at the Harvard T.H. Chan School of Public Health and Section Head for Epidemiology and Data Science at the Smith Center for Outcomes Research at Beth Israel Deaconess Medical Center. His research focuses on the design and analysis of randomized trials and observational studies, with an emphasis on causal inference and evidence synthesis. He also develops statistical methods that integrate diverse data sources to improve decision-making in clinical and public health settings.
Speaker Information
Issa Dahabreh, MD, ScD
Organizers
Related Events
ⓘ Harvard Chan School hosts a diverse array of speakers, invited to share both scholarly research and personal perspectives. They do not speak for the School, and hosting them does not imply endorsement of their views, organizations, or employers.