Ihsane Hmamouchi’s Post

sn-news: #ai #ml #llm #medical #health MedFuzz: Exploring the robustness of LLMs on medical challenge problems

A newly published study from IMO Health, Harvard Medical School, and Mass General Brigham displays AI’s potential for detecting cognitive decline early. Our study found that LLMs and traditional machine learning models trained on local EHR data will each produce unique errors. By combining these distinct models into one ensemble model, we can greatly enhance diagnosis precision. This is a major milestone in the #ClinicalAI space and our mission to transform healthcare by transforming data. Learn more here: https://bit.ly/3BTYb3e #HealthcareInnovation #AI #HealthIT

A newly published study from IMO Health, Harvard Medical School, and Mass General Brigham displays AI’s potential for detecting cognitive decline early. Our study found that LLMs and traditional machine learning models trained on local EHR data will each produce unique errors. By combining these distinct models into one ensemble model, we can greatly enhance diagnosis precision. This is a major milestone in the #ClinicalAI space and our mission to transform healthcare by transforming data. Learn more here: https://bit.ly/3BTYb3e #HealthcareInnovation #AI #HealthIT

This is fascinating and so incredibly powerful. A newly published study from IMO Health, Harvard Medical School, and Mass General Brigham displays AI’s potential for detecting cognitive decline early. Our study found that LLMs and traditional machine learning models trained on local EHR data will each produce unique errors. By combining these distinct models into one ensemble model, we can greatly enhance diagnosis precision. This is a major milestone in the hashtag #ClinicalAI space and our mission to transform healthcare by transforming data. Learn more here: https://bit.ly/3BTYb3e hashtag #HealthcareInnovation hashtag #AI hashtag #HealthIT

LLM shortfalls described well here in the case of medicine. Supplementary techniques, like RAG, etc need to be used to validate results, and create trustworthy answers. A medical example is well illustrated in this article. #AI #LLMs #AI4Good #MedicalAI #AIinMedicine https://lnkd.in/gQJFuTUf

Oooo, could AI soon replace doctors? The interesting thing is how this article describes the use of multiple LLMs together and how when there is uncertainty, it goes out to research the web to find more information. #accelerateinnovation

😷 "Why LLMs alone beat doctors, but doctors with LLMs don’t seem to improve."⚡️ Well, you would think physicians working with AI should lead to better results? Not always! A new study shows interesting psychological facts why this is the case 👇😉. 1) The #Eliza #Effect The Eliza effect refers to the tendency of users to over-trust and attribute human-like qualities to AI-generated insights, often leading them to overestimate the capabilities of AI. This effect can result in an unrealistic expectation of the AI's understanding and abilities, potentially causing users to rely too heavily on its outputs in various contexts. The caution and critical thinking of physicians (also found with laywers) may lead to disregarding potentially useful AI-driven insights. 2) Missing prompt engineering skills and time to learn. Prompt engineering becomes a barrier to the effective use of AI in clinical decision-making. LLMs such as OpenAI's o1 help with prompting with Chain of Thought (CoT) mechanisms. 3) The #AI #Mindset: To successfully implement AI, it is essential to address concerns and develop new formats in partnership with AI experts, along with clinical and healthcare professionals. Enabling physicians with user-centric tools and prompt engineering is important. 🔥 Follow me to learn more about Generative AI from a tech and business perspective 🪄👀 👋 Need help with enablement? Reach out to me to see support your physicians and doctors 🚀 😉 Article: https://lnkd.in/ea3cmWPR

This study highlights why identifying the appropriate use cases for application of AI in medicine is so important, as well as learning about the limitations and having realistic expectations. https://lnkd.in/euBRD3Ua #artificialintelligence #AIinMedicine #WellQAI #LLM #Clinical

The appropriate use cases for application of AI in medicine might not be shiny, but that is where we need to focus. #AIinMedicine #LLM #Artificialintelligence #WellQAI