Johan RBH Boie’s Post

The Doctor AI debate It’s still very early in the era of A.I. in medicine, especially with LLMs. Progress in medical A.I. won’t occur in a straight line. Doctor Eric Topol does a very good job summarizing some of the latest findings. One big step forward with a 16.000 patient study that showed a statistically significant 17% reduction of all-cause mortality (that is huge!). And then a review of 4 recent studies published that question whether we’re ready for LLMs in medical practice (and even coding). Conclusion? Promising, and much more work is needed.... https://lnkd.in/dNcnK4Nd

Google just released its latest paper on Med-Gemini, an LLM for Medicine. "Excellence in a wide variety of medical applications poses considerable challenges for AI, requiring advanced reasoning, access to up-to-date medical knowledge and understanding of complex multimodal data. Gemini models, with their strong general capabilities in multimodal and long-context reasoning, offer exciting possibilities in medicine. Med-Gemini’s performance suggests real-world utility by surpassing human experts on tasks such as medical text summarization and referral letter generation, alongside demonstrations of promising potential for multimodal medical dialogue, medical research and education. Taken together, our results offer compelling evidence for the promise of Med-Gemini in many areas of medicine, although further rigorous evaluation will be crucial before real-world deployment in this safety-critical domain". Source: https://lnkd.in/gjyYNNtF

Google just released its latest paper on Med-Gemini, an LLM for Medicine. "Excellence in a wide variety of medical applications poses considerable challenges for AI, requiring advanced reasoning, access to up-to-date medical knowledge and understanding of complex multimodal data. Gemini models, with their strong general capabilities in multimodal and long-context reasoning, offer exciting possibilities in medicine. Med-Gemini’s performance suggests real-world utility by surpassing human experts on tasks such as medical text summarization and referral letter generation, alongside demonstrations of promising potential for multimodal medical dialogue, medical research and education. Taken together, our results offer compelling evidence for the promise of Med-Gemini in many areas of medicine, although further rigorous evaluation will be crucial before real-world deployment in this safety-critical domain". Source: https://lnkd.in/gjyYNNtF

Garbage in, Garbage out. The major LLMs train on "generally available" data, but what data? Is it any good? Focused models with better input offer better results. #gigo #medical #healthcare #llm #ai

The Impact of Artificial Intelligence on Internal Medicine Physicians: A Survey of Procedural and Non-procedural Specialties - Cureus: The Impact of Artificial Intelligence on Internal Medicine Physicians: A Survey of Procedural and Non-procedural Specialties  Cureus http://dlvr.it/TD4qkx #ai #artificialintelligence

Google has just released their latest Multimodal LLM for the medical domain - Med-Gemini. This release boasts an impressive 91.1% accuracy on MedQA (USMLE) benchmark, and outperforms Med-PaLM 2 by 4.6%. On 7 multimodal benchmarks including NEJM Image Challenges and MMMU (health & medicine), Med-Gemini improves over GPT-4V by an average relative margin of 44.5%. More details can be found here -> https://lnkd.in/gAnTMPdC This type of work has the potential to open up access to world-class doctors for patients in remote parts of the world, as doctors and hospitals will be able to use such technology remotely to increase their efficiency.

#AI and #LLMs to generate physician notes to save them time and expense is unethical if at the same time it is passing information to the future (that's the purpose of medical records) that deteriorates patient outcomes. The #RCT is easy! Randomize half the physicians who only use old fashioned hand written notes and never use AI, and half to AI generated notes. Then count the outcomes. Q.E.D.

Stanford Medicine physicians have a new artificial intelligence tool to assist them when they message patients about test results. The technology drafts an interpretation of clinical test and lab results and explains them in a message using plain language, which a physician then reviews and approves. The technology adds to a growing number of AI-based tools poised to help physicians spend less time on administrative tasks and more time on more meaningful work, such as interacting with patients.

I'd make future hospital leaders play with games and simulations like this one. Chinese researchers developed "a simulacrum of hospital called Agent Hospital" that simulates the entire process of caring for patients. Patients, nurses, and doctors are autonomous agents powered by large language models (LLMs) and they developed it to enable a doctor agent to learn how to treat illnesses within the simulacrum. "𝑆𝑖𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑠 𝑠ℎ𝑜𝑤 𝑡ℎ𝑎𝑡 𝑡ℎ𝑒 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑜𝑓 𝑑𝑜𝑐𝑡𝑜𝑟 𝑎𝑔𝑒𝑛𝑡𝑠 𝑐𝑜𝑛𝑠𝑖𝑠𝑡𝑒𝑛𝑡𝑙𝑦 𝑖𝑚𝑝𝑟𝑜𝑣𝑒𝑠 𝑜𝑛 𝑣𝑎𝑟𝑖𝑜𝑢𝑠 𝑡𝑎𝑠𝑘𝑠. 𝑀𝑜𝑟𝑒 𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡𝑖𝑛𝑔𝑙𝑦, 𝑡ℎ𝑒 𝑘𝑛𝑜𝑤𝑙𝑒𝑑𝑔𝑒 𝑡ℎ𝑒 𝑑𝑜𝑐𝑡𝑜𝑟 𝑎𝑔𝑒𝑛𝑡𝑠 ℎ𝑎𝑣𝑒 𝑎𝑐𝑞𝑢𝑖𝑟𝑒𝑑 𝑖𝑛 𝐴𝑔𝑒𝑛𝑡 𝐻𝑜𝑠𝑝𝑖𝑡𝑎𝑙 𝑖𝑠 𝑎𝑝𝑝𝑙𝑖𝑐𝑎𝑏𝑙𝑒 𝑡𝑜 𝑟𝑒𝑎𝑙-𝑤𝑜𝑟𝑙𝑑 𝑚𝑒𝑑𝑖𝑐𝑎𝑟𝑒 𝑏𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘𝑠.𝐴𝑓𝑡𝑒𝑟 𝑡𝑟𝑒𝑎𝑡𝑖𝑛𝑔 𝑎𝑟𝑜𝑢𝑛𝑑 𝑡𝑒𝑛 𝑡ℎ𝑜𝑢𝑠𝑎𝑛𝑑 𝑝𝑎𝑡𝑖𝑒𝑛𝑡𝑠 (𝑟𝑒𝑎𝑙-𝑤𝑜𝑟𝑙𝑑 𝑑𝑜𝑐𝑡𝑜𝑟𝑠 𝑚𝑎𝑦 𝑡𝑎𝑘𝑒 𝑜𝑣𝑒𝑟 𝑡𝑤𝑜 𝑦𝑒𝑎𝑟𝑠), 𝑡ℎ𝑒 𝑒𝑣𝑜𝑙𝑣𝑒𝑑 𝑑𝑜𝑐𝑡𝑜𝑟 𝑎𝑔𝑒𝑛𝑡 𝑎𝑐ℎ𝑖𝑒𝑣𝑒𝑠 𝑎 𝑠𝑡𝑎𝑡𝑒-𝑜𝑓-𝑡ℎ𝑒-𝑎𝑟𝑡 𝑎𝑐𝑐𝑢𝑟𝑎𝑐𝑦 𝑜𝑓 𝟫𝟥.𝟢𝟨% 𝑜𝑛 𝑎 𝑠𝑢𝑏𝑠𝑒𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑀𝑒𝑑𝑄𝐴 𝑑𝑎𝑡𝑎𝑠𝑒𝑡 𝑡ℎ𝑎𝑡 𝑐𝑜𝑣𝑒𝑟𝑠 𝑚𝑎𝑗𝑜𝑟 𝑟𝑒𝑠𝑝𝑖𝑟𝑎𝑡𝑜𝑟𝑦 𝑑𝑖𝑠𝑒𝑎𝑠𝑒𝑠." They hope that this work could pave the way for advancing the applications of LLM-powered agent techniques in medical scenarios.In short, medical universities should start looking into similar ways of using AI to teach physicians through simulations how medicine works. https://lnkd.in/dciwSHnR

🚨 New Webinar Recording Available! 🚨 We kicked off 2025 with an insightful CoDEx webinar featuring Cornelius James, MD, a National Academy of Medicine Scholar in Diagnostic Excellence ('21-'22). Dr. James took us on a journey through the evolution of diagnostic AI, from early expert systems to today’s advanced machine-learning models. He also explored how medical education is adapting to ensure clinicians are prepared to use AI effectively in practice. One key takeaway? AI literacy is essential. As AI continues to shape diagnostic decision-making, healthcare professionals must be equipped to critically assess AI-driven recommendations—not as replacements, but as enhancements to clinical expertise. 🎥 Watch the full recording here: https://lnkd.in/gPfz6NSx