Name
Panel 2: Deploying generative AI systems in healthcare
Date & Time
Friday, September 20, 2024, 11:00 AM - 12:30 PM
Speakers
Evangelos Oikonomou, Yale School of Medicine Generative Ai In Cardiovascular Imaging: From Innovation To Real-world Implementation The advent of large language models and generative artificial intelligence (AI) has resulted in a paradigm shift in biomedical research. This talk will explore the applications and challenges of generative AI in cardiovascular imaging, with a focus on developing and deploying scalable and accessible solutions for cardiovascular disease diagnosis and prognosis.
Javier Elkin, ICRC Bridging Digital Health Inequities In Conflict
Mary-Anne "Annie" Hartley, Yale-Medicine Moove: Massive Online Open Validation And Evaluation For Continuous Community-guided Real-world Alignment Of Medical Chatbots Large language models (LLMs) have the potential to democratize access to medical knowledge. Unfortunately, the enormous potential of these models is either locked behind commercial/research licenses, limited in scale, or not generalizable to underserved populations. To address this issue, we developed Meditron, a suite of open-access LLMs co-designed with a global network of clinicians and specifically adapted to low-resource and humanitarian contexts. Meditron-3 is continuously pre-trained from Llama-3.1 [8B, 70B] and is currently the state-of-the-art fully open-source LLM for medicine according to the standard benchmarks. However, these benchmarks (mostly medical exam and multiple-choice questions), are static, generic, and have limited geographic scope. More importantly, they fail to adequately assess real-world clinical utility, safety, and equity. In this talk, I introduce the MOOVE: a Massive Online Open Validation and Evaluation platform that allows clinicians to collaboratively validate the real-world performance of LLMs in terms of helpfulness, harmlessness, bias, trust, and safety, while creating contextualized models for their own setting.
Lovedeep Dhingra, Yale School of Medicine Panel 2: Deploying Generative Ai Systems In Healthcare Panel Moderator
Aline Pedroso Camargos, Yale School of Medicine
Javier Elkin, ICRC Bridging Digital Health Inequities In Conflict
Mary-Anne "Annie" Hartley, Yale-Medicine Moove: Massive Online Open Validation And Evaluation For Continuous Community-guided Real-world Alignment Of Medical Chatbots Large language models (LLMs) have the potential to democratize access to medical knowledge. Unfortunately, the enormous potential of these models is either locked behind commercial/research licenses, limited in scale, or not generalizable to underserved populations. To address this issue, we developed Meditron, a suite of open-access LLMs co-designed with a global network of clinicians and specifically adapted to low-resource and humanitarian contexts. Meditron-3 is continuously pre-trained from Llama-3.1 [8B, 70B] and is currently the state-of-the-art fully open-source LLM for medicine according to the standard benchmarks. However, these benchmarks (mostly medical exam and multiple-choice questions), are static, generic, and have limited geographic scope. More importantly, they fail to adequately assess real-world clinical utility, safety, and equity. In this talk, I introduce the MOOVE: a Massive Online Open Validation and Evaluation platform that allows clinicians to collaboratively validate the real-world performance of LLMs in terms of helpfulness, harmlessness, bias, trust, and safety, while creating contextualized models for their own setting.
Lovedeep Dhingra, Yale School of Medicine Panel 2: Deploying Generative Ai Systems In Healthcare Panel Moderator
Aline Pedroso Camargos, Yale School of Medicine
Location Name
Kline Tower: 14th Floor
Full Address
Kline Tower - 14th Floor
219 Prospect St
New Haven, CT 06511
United States
219 Prospect St
New Haven, CT 06511
United States
Session Type
Workshop