Session (CTI)

PURPOSE: Advances in genAI are reshaping health economic analysis workflows, spanning academic methodological research through HTA submissions. As we increasingly engage with genAI-assisted modeling, clearer methodological guidance, performance evaluation standards, and implementation frameworks are needed. This workshop will present cross-stakeholder perspectives on: (1) recent genAI advancements and proposed frameworks for integrating agentic intelligence into economic modeling; (2) operational and methodological barriers to using genAI in industry and academic modeling workflows and the need for performance-based evaluation criteria; and (3) how HTA bodies may assess performance, transparency, and verification of genAI-assisted models. DESCRIPTION: Dr. Wang will contextualize recent advancements in genAI within HEOR and present her collaborative research on an implementation framework for integrating agentic intelligence into economic modeling (15 min). Mr. Feng will present the industry user’s perspective on integrating genAI into internal workflows, including how genAI-assisted health economic models should be evaluated on their performance rather than on their ability to replicate human-built models. He will discuss key barriers that limit modelers’ adoption of genAI tools in practice and will call for an initiative to establish a formal evaluation framework for genAI-assisted economic modeling (10 min). Dr. Wright will discuss the HTA perspective, highlighting stages where genAI-assisted approaches are likely to be most useful in evidence synthesis and modeling. She will also address the importance of transparency and validation in these approaches (10 min). Dr. Li will synthesize the discussion and present the academic user’s perspective, including practical concerns about the cost and access of specialized genAI tools, and methodological considerations related to ensuring comprehensive literature capture, transparent workflows, and reproducible modeling results. (10 min). Audience engagement will include polling questions and Q&A (15 min).