What the research found
Researchers evaluated the Epic Deterioration Index, a machine-learning early-warning tool, across 23,132 high-risk patients at 11 RWJBarnabas Health hospitals in New Jersey. After a multi-year rollout pairing the tool with staff training, alert-delivery redesign and automated rapid-response notifications, in-hospital mortality among high-risk patients fell from 23.1% to 18.6% — an 18% reduction in risk-adjusted odds of death — while rapid-response activations rose from 25.3% to 37.5% of stays, with no significant rise in ICU transfers.
A rare, well-governed counterexample to this tracker's usual pattern: the same vendor's Epic Sepsis Model missed two-thirds of cases in independent validation, while this Epic tool, backed by years of piloting, clinician training and continuous monitoring before go-live, delivered a real mortality reduction. The difference was implementation rigor, not the algorithm.