PS15: Advancing Precision Medicine through Statistics and ML/AI: Scientific Accomplishments of the ASA Precision Medicine Statistics Working Group

Precision medicine is redefining drug development by moving beyond one-size-fits-all approaches toward therapies tailored to individual biological characteristics. This paradigm shift relies on biomarkers, companion diagnostics (CDx), and the integration of real-world, omics, and imaging data as foundational elements of modern clinical development. At the same time, the growing use of ML/AI introduces new opportunities and complexities. Along with the adoption of ML/AI methods come substantial statistical and regulatory challenges that demand innovative solutions. The Precision Medicine Statistics Working Group (PMSWG) is a regulatory industry collaborative initiative established under the ASA Biopharmaceutical Section. Its mission is to develop best practices for biomarker and CDx development, promote harmonization of statistical methodologies, and address emerging challenges in precision medicine. The PMS Working Group brings together statisticians from industry and regulatory agencies to ensure that innovative analytics are grounded in rigorous statistical principles and aligned with regulatory expectations. This session will showcase the scientific accomplishments of PMSWG. It features two complementary presentations, one focused on advanced biomarker methodology and another on companion diagnostic development, followed by a regulatory-industry panel discussion. Together, these presentations will illustrate how modern statistics and ML/AI can be embedded into study design and analysis to improve patient selection, optimize evidence generation, and enable credible, explainable regulatory decision-making. Attendees will leave with: • Practical statistical strategies for predictive biomarker identification and cutoff selection, including uncertainty quantification and interpretability considerations. • A decision-focused framework for evaluating ML-enabled virtual twin and synthetic/external control approaches, including key assumptions and sensitivity analyses aligned with regulatory expectations. • Best practices and lessons learned for CDx co-development (analytical and clinical validation) and assembling an evidence package to support regulatory decision-making.