UPMC develops AI tool to predict mortality for patients facing hospital transfer

Listen
Text
  • Small
  • Medium
  • Large

UPMC researchers created a machine learning-powered tool that can rapidly predict mortality for patients facing a transfer between hospitals to obtain higher-acuity care, the Pittsburgh-based health system announced Feb. 8.

The research team used data from nearly 21,000 patients who were transferred to a UPMC hospital during a one-year period to create the tool, called SafeNET, short for "Safe Nonelective Emergent Transfers." 

After studying mortality risk tools currently used in intensive care units and admission settings, the researchers compiled a list of 70 independent variables used in one or more of these models, including vital signs, patient demographics and lab tests. They then examined UPMC billing data and inpatient EHRs to determine if those variables were recorded by the receiving hospital, focusing solely on those available within three hours of a patient's transfer. 

The list of variables was then cut down to 54, and eventually 14 to create the SafeNET algorithm, which can rapidly predict in-hospital, 30-day and 90-day mortality for patients at the time of transfer and deliver the results to physicians in less than five minutes. 

"Our overarching goal was to provide much-needed information to front-line physicians to trigger and inform shared decisions about the highest risk patients,” Daniel Hall, MD, one of the researchers, said in a news release. "This tool could help to direct additional resources to these patients to ensure that the plan of care is consistent with the patient’s values and goals. Frequently, transferred patients and their families don’t understand the severity of the illness they are facing and have unrealistic expectations about the outcomes that a transfer for higher-level care will produce."

Copyright © 2021 Becker's Healthcare. All Rights Reserved. Privacy Policy. Cookie Policy. Linking and Reprinting Policy.

 

Featured Whitepapers

Featured Webinars