Comment from Panoptive

Panoptive appreciates FDA’s initiative to explore AI-enabled real-time clinical trials. We strongly encourage FDA to include technical AI domain experts alongside clinical, regulatory, sponsor, CRO, and investigator stakeholders in the pilot’s design and evaluation. Real-time clinical trial AI should not be assessed only by whether it surfaces signals faster, but by whether the system is reliable for its intended use, governed by risk-proportionate controls, and tested against known failure cases such as false positives, false negatives, automation bias, hallucination, source-attribution failure, silent drift, and alert fatigue. FDA should require participants to map these failure modes in advance, document the controls used to mitigate them, and measure whether those controls are working in practice. Panoptive’s perspective combines clinical trial oversight infrastructure with deep AI evaluation and human-in-the-loop governance expertise. We recommend that FDA measure the full oversight loop: source event, AI-surfaced signal, qualified human review, documented sponsor or CRO decision, assigned action, and resolution. Real-time visibility only becomes real-time oversight when AI outputs are reliable, traceable, governed, and converted into accountable human decisions with inspection-ready records. FDA’s real-time clinical trials initiative is an important and timely step forward, and Panoptive would welcome the opportunity to support the pilot through technical input on AI governance, evaluation methodology, failure-mode mapping, risk-proportionate controls, and inspection-ready real-time oversight infrastructure. Youssef Kallel Founder & CEO Panoptive