
Creating Fair, Useful, and Reliable AI in Healthcare
Dr. Nigam Shah, Professor of Medicine at Stanford University and Chief Data Scientist for Stanford Health Care, lays out what it actually takes for an AI model to be useful in a hospital setting. He argues that accuracy alone is not enough, and walks through how model predictions translate into decisions, how clinicians act on those decisions, and where the chain breaks down in practice. The talk covers fairness problems that arise when training data underrepresents certain patient populations, reliability issues when models are deployed outside the conditions they were trained on, and the organizational work needed to make AI outputs genuinely actionable for doctors and nurses. Shah draws on his experience building and deploying predictive models within Stanford Health Care, using real examples from clinical workflows rather than abstract benchmarks. The webinar runs just under fifty minutes and is aimed at viewers with some familiarity with machine learning concepts.