Agentic AI in Evidence Submissions: Rigor, Trust, Traceability, & Compliance
Moderator
Xiaoyan Wang, PhD, Tulane University, New Orleans, LA, United States
Speakers
Ipek Ozer Stillman, MBA, MSc, Takeda, Cambridge, MA, United States; Gabriel Innes, VMD PhD, US Food and Drug Administration, Silver Spring, MD, United States; Wei Liu, MSc, PhD, J&J R&D, Potomac, MD, United States
Purpose Generative and agentic AI can substantially accelerate evidence generation and synthesis. Yet the development and adoption of Agentic AI-driven workflows for evidence submissions raises critical questions: what are the methodological and implementation considerations when automating documentation and evidence packages using agentic AI frameworks? How should industry redesign workflows to preserve trust, traceability, and compliance when AI becomes an active contributor to evidence development? How can organizations ensure transparency and methodological rigor? And what documentation pathways will payers and regulators expect as AI-generated evidence becomes more prevalent? This workshop brings together leaders from biopharma, regulatory science, and AI research to offer a practical, multidisciplinary view of this rapidly evolving landscape. Description The session will open with an overview of the evolving role of agentic AI in HEOR. Dr. Wang (Tulane/NouStarX) will present an implementation framework for agentic AI in evidence submissions, focusing on knowledge-base generation, documentation automation, and quality controls. Ms. Ozer-Stillman (Takeda) will share real-world experience applying AI-enabled processes to GVD, adaptations for local HTA submissions and value communications tools, highlighting practical considerations for auditability, traceability, and cross-functional alignment. Dr. Liu (J&J R&D) will discuss opportunities and constraints for AI-supported FDA engagements, including experiences related to programs such as Breakthrough Therapy Designation and Post-Marketing Requirements/Commitments, and will address the evidential standards and validation requirements necessary for AI-assisted evidence to be credible in regulatory interactions. Finally, Dr. Innes (FDA) will offer a forward-looking regulatory perspective on the transparency, robustness, and provenance expectations that may guide future acceptance of AI-generated evidence as organizations adopt agentic AI–driven workflows. The session will be followed by a moderated panel discussion focused on practical adoption guidance, governance considerations, and an open Q&A.
Topic
Health Policy & Regulatory, Health Technology Assessment, Methodological & Statistical Research