Post by Simon Williams

We hear it all the time - AI will replace clinicians. But does that hold true for surgeons and interventionists? I’m very proud to share our latest work just published in npj Digital Medicine, in which we set out to answer this question by comparing AI vs clinician performance in surgical and interventional video analysis. We screened nearly 40,000 studies, identifying those where an AI model built for surgical/interventional video analysis was directly compared with clinician performance. So thats tasks like using AI to identify a suspicious polyp on colonoscopy, point out critical anatomy in pituitary surgery, or label ‘safe’ areas for dissection during thyroid surgery. We found a few cool things: 1. AI alone was marginally better than clinicians alone. But wait - we then looked at how clinicians who used AI assistance performed when compared with clinicians who didn’t… 2. AI-assisted healthcare professionals performed better than unassisted healthcare professionals. Good news! But that includes all grades, from med students right up to grizzled old consultants/attendings. How does AI compare with true experts? 3. Senior clinicians using AI still outperformed senior clinicians who didn’t use AI. So what does this tell us? It echoes our group’s message that AI wont replace doctors; but *doctors who use AI might replace those who dont*… Finally, it tells us that the absolute gold standard is an expert using AI. A huge thanks to the team for this one - it was a mammoth piece of work, led tirelessly by our own Amir Rafati Fard. 👏 Full study link below.  https://lnkd.in/ebUjVmgR