Verifying AI in Medicine: Is Healthcare AI Actually Ready for Deployment?

Dr. Zak Kohane thinks that doctors are good at diagnosing common diseases. But doctors are often stumped by rare diseases, especially those that present atypically. There are 26,000 total diseases, and it's a complete fallacy to imagine a human being could be on top of the latest research and guidelines for all of them. Dr. Kohane was one of the first physicians in the field to recognize the potential for AI in diagnosis and treatment early on, even starting the first New England Journal of Medicine publication to explore the opportunities and pitfalls. But as recent headlines show, there's still much work to do to ensure we protect patients from harm and physicians from liability. 

Dr. Kohane joins us for a live-streamed webinar on October 13 to discuss whether AI is ready for the real world or if it should still live (for now) in the land of pilots. He's joined by Protege's Engy Ziedan, a career scientist and entrepreneur who believes that progress in AI will only be possible with the right training data. 

Expect us to cover:
● Why strong benchmark performance doesn’t always translate to real-world clinical readiness
● What healthcare AI should actually be compared against (we're personally not a fan of comparing AI to a panel of medical experts, which is just not how most patients experience care)
● Is the standard for AI too high if we expect it to be perfect? The status quo isn't perfect. Mistakes happen every day. 
● What evidence clinicians and health system leaders should expect before trusting AI in practice
● What health systems are doing to evaluate AI today, and what they should do tomorrow
● The opportunities to use AI in diagnosis and treatment, not just administrative use cases