Researchers have demonstrated the use of machine–human teamwork for reducing mortality in inpatient care across a large health system.
In this case, the one-two punch paired AI detection of dangerously declining patient condition with automated activation of a rapid-response squad with critical-care expertise.
The project was carried out at 11 RWJ Barnabas Health hospitals in New Jersey and is described in a study published July 29 in the New England Journal of Medicine’s NEJM AI.
For the study, Thomas Nahass, MD, MS, and colleagues at the health system and its academic affiliate, the Robert Wood Johnson Medical School at Rutgers University, compared the mortality outcomes of around 10,800 inpatients in a no-intervention cohort with outcomes of more than 12,300 post-intervention inpatients.
For the test AI system, they used the Epic Deterioration Index, or EDI, which runs continuously inside the EHR.
When watchful AI rings, healthcare professionals answer
Nahass and co-researchers found the smart software activated the response team at a 37.5% clip in the intervention group vs. 25.3% in the non-intervention group.
Almost half the intervention group, 46.6%, generated an alert to the rapid-response team.
Noting that not every alert spurred response-team activation, the authors report that risk-adjusted odds of inhospital mortality—after accounting for age, comorbidities, hospital type and other variables—were notably lower in the intervention cohort.
Specifically, deaths among high-risk patients fell from 23.1% to 18.6% following implementation of the AI-equipped early warning system
This represented an 18% reduction in the risk-adjusted odds of in-hospital death.
‘Right care at the right time’
In coverage of the project by the Rutgers news operation, Nahass says the aim of the technology is to identify at-risk patients before they reach a point at which recovery becomes harder.
“If we can get a critical-care eye on the patient sooner, we can change the course of their outcome,” Nahass adds.
Co-author Andy Anderson, MD, MBA, says the study demonstrates “how AI-enabled tools, when paired with experienced clinical teams, can help us identify patients at risk sooner and deliver the right care at the right time.” HealthExec
