Effectiveness of an AI–Enabled Intervention for Detecting Clinical Deterioration

In a cohort study conducted at a single academic center between January 17, 2021, and November 16, 2022, involving 9938 hospitalized patients, the implementation of an artificial intelligence (AI)-enabled intervention for detecting clinical deterioration was analyzed. This intervention, based on Epic Deterioration Index (EDI) score thresholds, aimed at reducing escalations in care, such as rapid response team activations, transfers to intensive care units, or occurrences of cardiopulmonary arrests. The study found a significant association between the AI-enabled intervention and a 10.4 percentage point reduction in the risk of care escalation during hospitalization. Utilizing a regression discontinuity design to enhance causal inference, these results underscore the potential efficacy of AI-driven interventions in improving patient outcomes by mitigating clinical deterioration risks in hospital settings.

Opens on Jama Network