As artificial intelligence becomes increasingly embedded in health systems, how can we know whether these tools are delivering meaningful health benefits? AI tools can be updated and redeployed faster than the evidence validating them, raising the question of whether current evaluation frameworks are built for the speed of adoption. This session explores what rigorous impact assessments for AI tools should address.

This session examines the state of AI in global health through a central question: What structural challenges does global health face, which of those can AI realistically help address, under what conditions, and where does hype outpace the evidence?

In this presentation, Dr. Goodarz Danaei will discuss emerging approaches to cardiovascular risk prediction and prevention, including the use of causal inference methods and nationally representative data to develop more accurate risk models. He will explore the limitations of existing prediction tools, which often rely on outdated cohort studies or nonrepresentative electronic medical record data, and present new research aimed at improving population-level prevention strategies in the United States and globally.