3 Comments

Very interesting! Is it possible to look at medical device safety using publicly available data on adverse events, FDA warning letters and safety communications?

Expand full comment
author

Hi Naveen, to some extent it is possible to use medical guidelines, FDA warning letters and other public data to check how much do AI models follow such recommendations. I'm working on another project on grounding AI models on medical guidelines - should be published here in a week at most. I would still assume that the biggest test is real world hospital data, if we want to use such tools in medical practice.

Expand full comment

Thank you Zeljko. I look forward to reading your post.

Expand full comment