NYU Langone Health said its artificial intelligence system PREVIA flagged all 30 patients later diagnosed with post-traumatic stress disorder (PTSD), as emergency departments face limits on routine screening.
The university’s Nov. 9 announcement says researchers analyzed accounts from 145 emergency department patients. In a comparison within that group, PREVIA flagged all 30 people who later developed PTSD. An emergency department clinician specializing in the disorder flagged 89% of those patients, NYU said.
NYU described PREVIA as a system that works from recordings of patients explaining why they sought emergency care. It combines clinical risk signs with protective factors, such as coping skills and social support, to produce an assessment with an explanation. The announcement says researchers compared the system’s results with diagnoses based on the Clinician-Administered PTSD Scale for DSM-5.
The announcement did not report how often PREVIA flagged patients who did not later develop PTSD, so its false-positive rate cannot be determined from the comparison. NYU separately said the system identified more than 95.4% of patients considered at risk, but did not define that performance measure or say whether testing used a separate patient group.
About 28% of emergency departments routinely screen for PTSD, according to the announcement. It says roughly two-thirds of those departments use validated tools, and cites overcrowding and appointments that typically last 15 to 20 minutes as barriers to broader screening.
NYU said the first month after trauma is an important period for identifying risk because many patients have no further contact with the hospital. Senior study author Katharina Schultebraucks, an associate professor of psychiatry and population health at NYU Langone, said PREVIA “can serve as an AI screener that identifies at-risk patients and explains why.”
NYU said the study was published online Nov. 9 in Nature Communications.
