Our piece is out in NEJM AI! More than 1,000 AI-enabled medical devices have been cleared by the FDA, but most are still evaluated using old data from a handful of hospitals. Unfortunately, these tools often fail when deployed in other clinics or as they see newer patient cases. And when they do, there's no alarm...just a confident wrong answer. We thus need a solution that protects patients without slowing innovation. In our new paper, we propose that the FDA require: - before clearance: multisite validation that reflects the diversity of our patients - after deployment: continuous monitoring so issues get flagged and addressed quickly Endlessly grateful for guidance and insights from Dr. Roxana Daneshjou, Dr. Kavita Patel, star student Alia Sajjadian, and the editorial team at NEJM AI. 📄: https://lnkd.in/gGX-giHi *Happy to send a PDF, please dm or comment below
Ank Agarwal such an important paper and topic, thank you for highlighting this critical patient safety and AI accountability and liability gap. There is so much focus on pre deployment validation and clearance when the majority of focus should be on post deployment when it is most crucial. AI governance in regulated med device should be throughout the life cycle and continuously monitored, not just at model level but at runtime, our work is focused on the evidenxe infrastructure for this. Most of the FDA and MHRA auth med device companies we have spoken with have expressed need/desire for clearer guidance and requirements on what is required from regulatory bodies, until this happens, there is no motivation or urgency for companies to figure out how. The overall feeling is nobody is really asking for this evidence now, so it isn't a priority yet. Your work helps move the needle. I would love the full pdf please!
Great perspective! I think if we could agree on appropriate post-market surveillance, we could even REDUCE the upfront regulatory burden for low-risk devices
I’d love a PDF copy please, very interested in this!
Great observation Ank - Congratulatiobs on thus publication!! Pkease send pdf👍
Congrats on the publication 👏 please send it to me
Congrats 🎉 would love to read the pdf
Incredible!
Insightful work Ank!
Congrats Ank!!
Post-market surveillance already exists for drugs and devices. Extending that logic to AI is less a new regulatory burden than finishing an unfinished one. That framing might move this faster than treating it as novel.