Newsletters
Reducing Errors, Detecting the Rare: How GenAI is Transforming Care
2025-08-16 00:00
OpenAI and Penda Health’s study of 40,000 patient visits showed that a ChatGPT-powered copilot reduced diagnostic errors by 16% and treatment errors by 13% in Kenyan clinics, with even greater impact on safety-critical cases. At the same time, our Ge...
Integration of GenAI in Medical Education and Clinical Training
2025-07-10 00:00
GenAI, powered by rapidly advancing large language models, is transforming medical education and clinical training. Leading institutions like Harvard and Stanford are integrating GenAI into curricula, enabling new approaches to clinical reasoning, si...
GenAI-Based Monitoring of Cancer Care Quality Wins Global Challenge
2025-06-06 00:00
Our GenAI-ELHS solution for monitoring cancer care quality was selected as a winner in the global Transforming Cancer Navigation with Open Data & APIs Challenge. This tool empowers patient navigators and clinicians—especially in low-resource settings...
Nature Publishes First Autonomous AI Doctor Study
2025-05-05 00:00
Google’s AMIE became the first autonomous AI diagnostic agent rigorously evaluated and published in Nature, outperforming human doctors in diagnostic accuracy and clinical communication. Built on Google’s Med-PaLM model, AMIE’s success highlights a m...
More Progress in Open-Source LLMs and GenAI Agents for Medical Diagnosis
2025-04-05 00:00
Major studies in Nature, JAMA, and NEJM highlight rapid progress in GenAI for diagnosis. GPT-4 boosts physician accuracy; open-source Llama 3.1 shows near-proprietary performance. A multi-agent GPT-4 system further improves diagnostic reasoning, whil...
The Future of Healthcare with GenAI – Insights from Dr. Isaac Kohane
2025-03-05 00:00
Generative AI (GenAI) is set to transform healthcare. Dr. Isaac Kohane, Chair of Biomedical Informatics at Harvard, advocates for using GenAI to empower physician assistants and nurse practitioners in primary care. He also highlights the need for a d...
DeepSeek Facing Big Challenges Ahead
2025-02-04 00:00
DeepSeek shook the GenAI world, claiming to develop its top model for
Synthetic patient data – key to accelerating responsible LLMs in healthcare
2025-01-05 00:00
LLM development will soon run out of data. Synthetic data will become a key enabler for LLMs, particularly in healthcare, where most doctors are facing a bottleneck: no data, no LLMs to use for clinical research. However, the ELHS Institute has recen...
New Idea for FDA’s Life Cycle Regulation of Open-Source LLMs
2024-12-03 12:00
My new idea for FDA’s life cycle regulation for the non-SaMD: Allowing voluntary submission of validation data of fine-tuned open-source LLMs to help doctors to choose LLMs for clinical research and applications. ELHS Institute starts to release vali...
GenAI Wins Nobel Prize, Yet LLMs Still Lack Clinical Evaluation
2024-11-02 00:00
Research on artificial neural networks has won the Nobel Prize, marking the first gift to the new era of GenAI. However, healthcare-focused GenAI LLMs still lack proper clinical evaluation, as highlighted by the FDA Commissioner. AJ and ChatGPT discu...