Bayesian Dynamic Power Prior for Survival Analysis: Optimizing External Control Data Integration

Sanhita Sengupta

Sanhita has a PhD in Statistics from University of Minnesota with focus in causal inference and machine learning. She currently works at Bristol Myers Squibb as Sr. Manager, Data Sciences. Her current areas of research are dynamic borrowing, covariate adjustment, prognostic modeling using multimodal data, and real-world evidence.

Abstract: Incorporating historical control data to supplement small control arms in randomized trials can enhance efficiency but raises concerns about bias. To address this, we extend Bayesian dynamic borrowing with power priors to survival analysis, leveraging hazard ratios as the primary estimand. Our novel approach integrates empirical Bayes estimation to determine the optimal borrowing strength, using the asymptotic normal distribution of the log-hazard ratio between external and internal controls. For inference, we combine Bayesian bootstrap, empirical Bayes estimation, covariate adjustment, and multiple imputation to comprehensively capture uncertainty. Through simulation studies and application to real-world data from CheckMate-057 (advanced non-squamous NSCLC), we demonstrate the method’s robustness. We also introduce an efficient algorithm for variance estimation, ensuring computational feasibility even with multiple imputation. This session will highlight practical insights into applying Bayesian borrowing for real-world evidence integration in clinical trials, offering a rigorous yet adaptable framework for regulatory-compliant trial design.

Enhancing Clinical Trials with Generative AI: Reducing Burden, Expanding Reach, and Driving Equity

Erin Reynolds

Erin Reynolds brings over 10 years of experience as the lead for Business & Trial Diversity Analytics at AbbVie. Erin’s primary purpose is to partner closely with study teams and functional area leaders to use clinical, operational, and real-world data to transform drug development while improving clinical trial operations and access for both patients and sites. Her recent work has focused on AbbVie’s diversity and patient inclusion priorities and clinical site centricity efforts toward addressing patient and site burden, increasing access and awareness, and ensuring overall clinical trial representativeness. Erin earned her Bachelor’s in Biology from Loyola University Chicago and her Master's in Applied Probability & Statistics from Northern Illinois University.

Abstract: During this presentation, we’ll look at how AbbVie is taking action to reduce site and patient burden, while also expanding our patient reach by working with new and new to research sites and physicians. By harnessing Generative AI, AbbVie has enhanced the protocol planning process by synthesizing unstructured data gathered from the insights and perspectives of both sites and patients. This approach elevates the site and patient voice, ensuring that their needs, preferences, and experiences are considered in the development of clinical trial protocols. And in an effort to increase equitable access to clinical trials, AbbVie explores the benefits of working with new to AbbVie and new to research sites and physicians. Performance metrics will be shared, highlighting the various factors that are considered: screening, enrollment, protocol compliance, and risk-based quality management signals.

Unlocking the Promise of RBQM with data and AI

Pranava Goundan and Sharma Ramanathan

Pranava Goundan is a Principal at ZS, where he leads the life sciences R&D AI practice. Pranava has extensive experience working with biopharma and biotech clients on clinical trial design and optimization, clinical trial feasibility & planning, clinical operations, biometrics, real world evidence, AI-driven clinical decision support, medical affairs and value & access. Pranava also advises clients on the rapidly evolving AI landscape and approaches to harnessing advances in Generative AI, foundation models, and AI agents to drive clinical trial acceleration. Pranava holds a Ph.D. in Operations Research from MIT and a bachelor’s degree in Computer Science from the Indian Institute of Technology, Madras.

Sharma Ramanathan is a Principal at ZS in the life sciences R&D Excellence practice. With over 15 years of experience in life sciences, Sharma helps several pharma clients optimize clinical trials end to end through data science, big data technologies for data management and governance, real world evidence and other advisory and consulting services. Sharma holds a PhD in Global Health from the Northwestern University Feinberg School of Medicine, a degree in information science from Stanford, and an MBA from the SP Jain School of Global Management.

Abstract: In the ever-evolving landscape of clinical trials, Risk-Based Quality Management (RBQM) has emerged as a transformative approach, promising substantial improvements across the drug development lifecycle, from study design to submission. Recent industry experience underscores the real-life impact of this approach, in marked improvements in significant protocol deviations, on-site monitoring time, and in Key Risk Indicator (KRI) metrics at clinical sites. These benefits are projected to expand as more studies adopt integrated RBQM practices. Join us to explore the detailed narrative of implementing a data and AI-powered RBQM ecosystem to accelerate clinical trials, reduce costs, and maintain uncompromising quality.