Using AI to Detect Hospital Patient Complications

Suchi Saria, founder and CEO of Bayesian Health and professor at Johns Hopkins, discusses the challenge of using artificial intelligence (AI) to detect when hospital patients are at risk of potentially deadly complications. The process involves collecting detailed patient data, understanding the clinical processes involved, and building and refining AI models. Saria's company has focused on detecting sepsis, a life-threatening condition, as well as other conditions like pressure ulcers. Studies have shown that the AI system can detect complications earlier, improve treatment timing, and lead to better patient outcomes. The challenge lies in building trust with clinicians and integrating the AI system into existing healthcare practices.