A study published in NEJM AI found that an AI-driven early warning system reduced deaths among high-risk hospital patients from 23.1% to 18.6% across 11 RWJBarnabas Health hospitals. The research, led by Dr. Thomas Nahass, evaluated outcomes for 23,132 patients and tracked how the Epic Deterioration Index (EDI) flagged clinical decline faster than bedside observation alone.
Nahass, vice president of Health Informatics and an intensive care physician at RWJBarnabas Health, said clinicians were initially skeptical. They would assess a patient as stable, only to see that patient transferred to the ICU two days later. "There's something I'm not seeing clinically that the algorithm is picking up," Nahass said, describing the moment staff recognized the tool's value.
The EDI continuously analyzes data already captured in the electronic health record - vital signs, lab results, nursing assessments, and age. It recalculates risk scores every 15 minutes. When a patient enters the highest-risk category, the system automatically alerts the rapid response team.
How the alert system works in practice
Scores appear in multiple locations within the patient chart. A yellow indicator signals scores between 31 and 59. Red marks scores above that threshold. A red flag also appears on the left side of the screen for high-risk patients. These visual cues aim to make deterioration status immediately obvious during routine chart review.
Rapid response team activations among high-risk patients rose from 25.3% of hospital stays to 37.5% after implementation. Critical care specialists then assess the patient to determine whether additional interventions are needed. Transfers to intensive care units did not significantly increase, even as mortality rates dropped.
Rollout and systemwide integration
The tool was first integrated into the electronic health records at Robert Wood Johnson University Hospital, the system's academic medical center, in early 2023. After researchers observed improved outcomes, RWJBarnabas Health expanded the EDI to its other hospitals in late 2023 and through 2024. Epic makes the Deterioration Index available to any customer, and health systems can use Epic's model or develop their own.
Nahass stressed that the algorithm is only one part of the equation. Workflow implementation carries equal weight. Clinicians must be engaged in a way that assures them AI is not taking away their autonomy. Teams need to know they can deliver the care they want without being beholden to a system, he said. For hospitals considering adoption, "it's really the integration," Nahass said.
Why this matters for healthcare professionals
The study demonstrates that an AI early warning system can surface clinical deterioration signals that experienced clinicians may not detect at the bedside. For nurses, rapid response teams, and critical care specialists, the tool shifts intervention earlier - before a patient reaches a point where treatment becomes far more difficult. The finding that ICU transfers did not spike alongside increased rapid response activations suggests the system helps target resources toward patients who genuinely need escalation, rather than generating false alarms that overwhelm staff. Understanding how these scores integrate into existing chart workflows is becoming part of the clinical skill set, especially as health systems increasingly embed AI for Healthcare into electronic health records. For professionals handling documentation, AI for Medical Records Clerks training can clarify how risk scores draw from the same records they manage daily.
Your membership also unlocks:
