Toward Virtual Patient: AI Accelerating Medical Discovery
Speaker: Hoifung Poon, PhD
Affiliation: Recursion AI; previously Microsoft Research
Date: September 2026
Watch the recording on YouTube
Can AI help create a “virtual patient” and transform how new treatments are discovered? Hoifung Poon, PhD, explores how multimodal generative AI could accelerate medical discovery through digital twins and virtual clinical trials.
The work presented in this talk was developed during Dr. Poon’s tenure at Microsoft Research. He subsequently joined Recursion as its first Chief AI Officer.
Abstract
Human biology stems from four billion years of evolution, which creates a proliferation of complexity and heterogeneity. The combinatorial explosion of biological hypotheses easily surpasses the number of atoms in the universe. Meanwhile, biomedical discovery advances at the snail pace of clinical trials, each taking years and costing over $100 million. As we enter the era of precision health and try to tailor treatments for each individual, continuing on today’s discovery processes is clearly unsustainable.
Our overarching research agenda lies in advancing multimodal generative AI to create a biomedical world model by bridging virtual patient and virtual cell, thus turning real-world patient journeys into a discovery engine through a generalized lab-in-the-loop flywheel. This enables us to synthesize population-scale real-world evidence from hundreds of millions of patients and accelerate biomedical discovery through AI-powered virtual clinical trials.
This in turn fuels diverse applications from imminent productivity gains via automating digitized knowledge work, to moonshot creativity gains via creating biomedical wind tunnels for in silico simulation. By leveraging AI to close the loop across clinical care and frontier research, we have a clear line of sight toward scalable drug discovery.
Speaker bio
Hoifung Poon is a globally recognized leader in biomedical AI, working at the intersection of frontier AI research and real-world applications in biology and medicine. He joins Recursion as its first Chief AI Officer, responsible for the company’s end-to-end AI strategy and execution across drug discovery and clinical development.
Prior to Recursion, he spent 15 years at Microsoft, most recently as General Manager at Microsoft Research, where he helped create Microsoft Health Futures, built a founding team on biomedical AI, and launched the real-world evidence mission for precision health. His research centers on developing a unified biomedical world model by bridging virtual patient and virtual cell, thus turning real-world data into a discovery engine through a generalized lab-in-the-loop flywheel.
His team and collaborators are among the first to explore large language models (LLMs) and multimodal generative AI in biomedical applications, producing popular open-source foundation models such as PubMedBERT, BioGPT, BiomedCLIP, LLaVA-Med, and BiomedParse, with tens of millions of downloads. His latest publications in Nature and Cell feature digital pathology and spatial proteomics foundation models such as GigaPath and GigaTIME.
He has led successful research partnerships with large health providers and life science companies, creating AI systems in daily use for applications such as molecular tumor boards and clinical trial matching. His prior work has been recognized with Best Paper Awards from AI venues including NAACL, EMNLP, and UAI, and he was named the “Technology Champion” by the Puget Sound Business Journal in its 2024 Health Care Leadership Awards.
He received his PhD in Computer Science and Engineering from the University of Washington, specializing in machine learning and natural language processing. He is an Affiliate Professor at the University of Washington School of Medicine.
Rights and attribution
This recording and accompanying text are presented for educational access with attribution to the speaker. Speaker views are their own. No open license is implied unless explicitly stated.
This is not an official Fred Hutch page; all views expressed here belong to the individual expressing them.