AI is rapidly entering biometrics workflows — but what actually holds up under FDA review, and where do teams most often go wrong?
Hanrui Zhang — DHT Review Lead, U.S. Food and Drug Administration (FDA)
Yali Li — Vice President, Biostatistics & Data Management, bioMérieux
Ran He — Founder, THC Lawyers
Louise Liu — CEO, Hill Research (Moderator)
In this episode of AI on the Hill, we bring together perspectives from the U.S. FDA, global biostatistics leadership, and legal experts to unpack how AI is being evaluated in regulated clinical development today, beyond hype, demos, or exploratory use cases.
The conversation dives into the real pitfalls sponsors face when applying AI to biometrics, including bias risks, validation gaps, governance challenges, and the blurred line between decision support and submission-grade evidence. Panelists discuss why AI is not yet ready to serve as a clinical endpoint, how AI-generated analyses may (or may not) complement pivotal trials, and what FDA reviewers actually look for when AI influences statistical analysis or decision-making.
-What differentiates exploratory AI from regulatory-acceptable evidence
-Common biometrics and subgroup bias pitfalls that raise red flags during FDA review
-The role of human-in-the-loop, documentation, and auditability in AI-enabled workflows
- When AI adds credibility, and when it increases regulatory and legal risk
- How teams should prepare for AI accountability as adoption accelerates
This episode is essential listening for professionals in biostatistics, biometrics, clinical development, regulatory affairs, and AI-enabled analytics who want a clear, regulator-aware view of where AI truly fits, and where caution is still required.
Recorded live | Public Zoom Webinar
Subscribe to AI on the Hill for expert conversations at the intersection of AI, biometrics, and clinical development.