AI-Powered Early Warning System to Improve Patient Safety
1:30pm - 2:30pmThursday, March 8
Las Vegas - Venetian Convention Center
Early detection of physiologic deterioration in order to reduce in-hospital mortality and prevent unplanned transfers to the intensive care unit (ICU) is a National Patient Safety Goal. Utility of non-automated early warnings system (MEWS) and Rapid Response Teams (RRT) to reduce in-hospital mortality are often limited due to inadequate and inconsistent alerting mechanisms. We describe the development, validation, and implementation of an automated, real-time AI-Powered Early Warning System (EWS) model for predicting risk of unplanned ICU transfers or cardiopulmonary arrests outside the ICU. The automated model outperforms non-automated models and unaided clinician observation, leading to improved care and patient safety. Post-deployment evaluation led to adoption of novel features: 1) alerts delivered with contextual reasons and triggers to facilitate rapid targeted clinical assessment by end-users, and 2) enhanced filtering to improve targeting of patients amenable to intervention.