Powering Proactive Healthcare: Leveraging AI to Reshape Our Healthcare System

Many of us know that AI has tremendous potential to change healthcare. But we are still far from leveraging its full potential. This talk presents a new model, MetisAI. MetisAI leverages large datasets to accurately find millions of patients at risk of adverse health events. This allows timely medical interventions that reduce costs while improving lives.

MetisAI is a transformer model capable of producing a wide spectrum of individual-level predictions that address clinical risk, healthcare cost, and even the timing of future clinical events. Historically, these predictive capabilities would require building dozens of separate models. By leveraging the latest research into transformer models, we have built a single foundation model that understands events in healthcare timelines at the individual level. Our model’s predictions go beyond the typical risk score. MetisAI is able to provide straightforward predictions both of the likelihood of a medical event and its timing. For example, we predict who will end up in the emergency department, which conditions will drive them to seek care (e.g. anxiety, depression, heart failure, etc.), and when they will end up there (e.g. in the next 7-9 months). These predictions allow health systems and health plans to take preventative action.

MetisAI has already been successfully integrated into existing workflows by healthcare providers and health plans to surface rising-risk members for proactive intervention. The talk will review results from some of these real-world case studies. Attendees will come away with an understanding of how AI foundation models are not just theoretical innovations but practical tools actively reshaping healthcare delivery.