PS12: Digital Twins in Clinical Development: Methodology and Regulatory Landscape

Digital twins, predictive models that estimate counterfactual patient outcomes using baseline characteristics, offer potential to improve efficiency across clinical development. Applications include virtual control arms to reduce sample sizes and patient enrichment to improve trial success rates. This session explores methodological foundations, implementation challenges, and the evolving regulatory landscape. Topics include: (1) statistical and machine learning methods for outcome prediction and treatment effect estimation, (2) regulatory qualification pathways including EMA's PROCOVA, (3) validation frameworks and evidentiary standards, and (4) practical lessons from early implementations. The session features speakers from industry, FDA, and academia to provide complementary perspectives on appropriate use and validation of these methods in drug development.