AI Optimization Subcore Office Hours and Discussion Forum

Discussion Overview
Evidence-based medicine has the potential to transform biomedical knowledge into better clinical decisions, but only when existing evidence can be fully utilized and new evidence can be efficiently generated. Dr. Qiao Jin's research tackles these bottlenecks by addressing two connected problems. First, how can we realize the full value of existing biomedical evidence for real-world information needs? Second, how can AI accelerate the generation of new biomedical evidence? His projects formulate medically important applications into well-defined AI tasks and build rigorous evaluation frameworks, such as PubMedQA. Together, this work aims to facilitate evidence-based medicine by using AI to help people use the evidence we already have, generate the evidence we still need, and develop the standards to evaluate medical AI in practice.

About the Speaker
Dr. Qiao Jin is a Research Fellow at the US National Library of Medicine, supported by the NIH K99/R00 Pathway to Independence Award. He received his BS and MD degrees from Tsinghua University. Dr. Jin works on AI for Evidence-Based Medicine and has published more than 50 peer-reviewed articles with over 7,500 citations. Research he has led or co-led, including PubMedQA and TrialGPT, has appeared in Nature Methods, Nature Communications, npj Digital Medicine, eBioMedicine, NeurIPS, ACL, EMNLP, and SIGIR. Dr. Jin developed the MedCPT foundation models, which have been downloaded over 5 million times and power NLM products that serve millions of users. His work has also been adopted by Google, Microsoft & OpenAI, Meta, Anthropic, and Nvidia, and featured by NIH News Releases, NIH Catalyst, POLITICO, Nature, Nature Biotechnology, and Medscape. Dr. Jin serves as Associate Editor of JMIR, Area Chair for ACL Rolling Review and NeurIPS, and Editorial Board Member for JAMIA.

Related links
AI Optimization Subcore Webpage