Survival Analysis in the Era of Generative AI

Purpose Survival analysis, and parametric survival curve selection in particular, is a critical yet challenging component in health economic modeling, often introducing significant uncertainty in cost-effectiveness analyses. This panel discusses the application of Generative AI (GenAI) to support human decision making, by comprehensively assessing both trial and external data sources. The session aims to bridge the technical innovation in GenAI with practical applications in health economics, presenting evidence-based insights and exploring the implications for HTA and industry practices. Description The session begins with an introduction by moderator Sven Klijn (10’) to the challenges of parametric survival curve selection and the potential of GenAI to address these limitations. Will Rawlinson from Estima Scientific will discuss the technical innovation in GenAI (15’), covering architecture and methodology of the GenAI system, including how it processes complex trial data alongside external sources to recommend optimal curve selections. He will also discuss the role of the human and implementation aspects of the Human/AI feedback loop. Prof. Nick Latimer (University of Sheffield and Petauri Evidence), a world-leading expert in survival analysis, will critically evaluate the results that can be achieved through this approach (15’), highlighting comparative advantages over conventional human-only methods while addressing potential limitations. Siguroli Teitsson (BMS) will conduct a broader examination of implications for survival analysis for HTA, practical implementation considerations, and future directions for this technology (10’). Following these presentations, the session concludes with an interactive audience discussion, addressing key concerns regarding transparency, reliability, and clinical plausibility of AI-assisted curve selection methodologies. This session will be particularly valuable for health economists, outcomes researchers, HTA bodies, and industry professionals seeking to understand how GenAI can be responsibly integrated into established methodological frameworks to enhance decision-making quality.