Microsoft Research revisits its book ´The AI revolution in medicine´ in a podcast episode that pairs peter lee with two senior health leaders: dr. Umair Shah, former Washington state secretary of health and founder of Rickshaw Health, and dr. Gianrico Farrugia, president and CEO of Mayo Clinic. The conversation maps the practical and strategic consequences of generative and predictive artificial intelligence across public health, healthcare delivery, research and medical education. It is both a stock-taking exercise and a blueprint for where leaders are placing bets.
Shah frames public health as the invisible ´offensive line´ that enables clinical care to succeed and argues it has been chronically underfunded and misunderstood. He describes pandemic-era operational gaps—fragmented data pipelines, faxed reports, mismatched patient records, supply chain blind spots for personal protective equipment—and shows where artificial intelligence could help: cleaning and stratifying messy surveillance data, accelerating epidemiological analysis, tailoring public communications, and narrowing the divide between community health and clinical response. He warns that if public health practitioners are not at the table when tools are designed, technology will be done ´to´ and ´for´ them rather than ´with´ them, and that responsible deployment must guard against phantom information and bias that could deepen inequities.
Farrugia recounts Mayo Clinic´s multi-year strategy that led to the Mayo Clinic Platform, begun around 2018, which centralizes de-identified longitudinal data to power applied models. He emphasizes a decentralised adoption model, clinical ownership, and rigorous validation. Mayo now runs hundreds of algorithms in practice, has programs such as CEDAR for early detection of patient deterioration, and has digitized pathology at scale to build foundation models like Atlas. Generative models have accelerated tasks from radiology report generation to genomics and enabled new pathways for home-based monitoring and treatment.
Both guests converge on governance and education: validation, stewardship, and embedding AI literacy in training are essential. The episode positions artificial intelligence as a tool that can shift systems from reactive to proactive, while stressing that technology must be coupled with domain expertise, ethical oversight, and public health engagement to deliver equitable health gains.
