PS22: Harnessing Real-World Evidence (RWE) in Regulatory Submissions: Advances, Methodologies, and Practical Applications

Real world evidence (RWE) continues to play an expanding role in regulatory decision making, driven by evolving FDA guidance, advances in causal inference methodology, and growing adoption of AI/ML techniques. This parallel session brings together perspectives from FDA and industry to examine how RWE is being responsibly and rigorously integrated into regulatory submissions, highlighting both methodological innovation and practical experience. The session will open with a regulatory perspective from Pallavi Mishra Kalyani (FDA), who will discuss current FDA thinking on the use of RWE in regulatory decision making, including emerging expectations for data quality, analytic rigor, and transparency. Her presentation will set the stage for understanding how recent regulatory developments inform the design, analysis, and interpretation of RWE studies submitted to FDA. Next, Tim Smith (Takeda) will present a series of real world case studies illustrating how AI and RWE are being applied across rare, chronic, and complex disease settings. Drawing from industry–academic collaborations, his talk will highlight both the promise and the practical challenges of deploying machine learning, computer vision, and large language models within RWE research—emphasizing issues such as data quality, interpretability, validation, and operational barriers that affect real world applicability and regulatory relevance. The session will conclude with a methodological deep dive from Hana Lee (FDA), who will provide a landscape overview of Targeted Maximum Likelihood Estimation (TMLE) for regulatory purposes. Her presentation will review how TMLE has been applied in FDA submissions and within FDA CDER regulatory science initiatives, illustrating how advanced causal inference methods—combined with flexible machine learning—can support robust and transparent regulatory analyses. Following the individual presentations, the session will feature an interactive panel Q&A with all three speakers, joined by additional panelists Mark van der Lann and Weili He. The panel discussion will allow for deeper dialogue on methodological best practices, regulatory expectations, AI/ML integration, and lessons learned from real world regulatory experiences, offering attendees practical insights they can apply to future RWE studies. Collectively, this session provides an integrated view of regulatory guidance, cutting edge methodology, and real world application, equipping attendees with a forward looking understanding of how RWE and AI can be used to support high quality regulatory submissions and sound decision making.