Predictive AI Evaluation and Monitoring

January 13th, 2026

AI prediction models are widely viewed as powerful tools to improve healthcare, but rigorous evidence of their impact in practice remains sparse. In this talk, Dr. Robert Gallo will describe the deployment of a deterioration prediction model at Stanford Health Care and how the model was evaluated using a quasi-experimental regression discontinuity design. He will then show how similar approaches can be used for a key emerging challenge in medical AI: monitoring model performance and safety over time in real-world care.

Robert Gallo, MD

Dr. Robert Gallo is a medical informatics research fellow in the Department of Health Policy and the VA Palo Alto Health Care System’s Center for Innovation to Implementation. He obtained his medical degree at Washington University School of Medicine and subsequently completed his residency training in Internal Medicine at Stanford. Dr. Gallo’s research focuses on inpatient health services delivery, particularly for diabetes and cardiovascular disease. He also has interest in the evaluation and implementation of prediction models