The Pulse

NYU Langone’s PREVIA Flags All 30 Later PTSD Cases in a Head-to-Head Test

NYU Langone Health researchers report that PREVIA flagged all 30 cases in a patient group that later developed PTSD. The AI tool analyzes emergency-room narratives to assess risk and explain its reasoning.

AI.info Team · Written with AI · fact-checked · 2026-11-09

PREVIA flagged 100 percent of 30 cases from the patient group that went on to develop PTSD; an emergency-department clinician specializing in PTSD correctly identified 89 percent of those cases. The set of AI models, each analyzing trauma narratives from 145 emergency-department patients, correctly identified more than 95.4 percent of at-risk patients. The study was published online November 9 in Nature Communications.

PREVIA’s 100% result came from 30 later PTSD cases

The patients described what brought them to the emergency department, and their answers were recorded and transcribed for analysis. The researchers examined whether those accounts could help identify who would develop post-traumatic stress disorder a month later.

Senior study author Katharina Schultebraucks, PhD, associate professor in the departments of psychiatry and population health at NYU Langone, said:

“PREVIA can serve as an AI screener that identifies at-risk patients and explains why, without adding any time to intake in busy emergency rooms.”

Katharina Schultebraucks, PhD, associate professor in the departments of psychiatry and population health at NYU Langone

From trauma narrative to an explained risk score

PREVIA combines AI tools selected for different strengths: the Llama model tends to emphasize signs of PTSD risk, while Mistral highlights protective and contextual factors. Its first layer scans for symptoms during the initial trauma, such as loss of control or amnesia, and early symptoms in the emergency department, such as feeling disconnected from the body.

A second layer weighs risk factors including pre-existing mental health problems and lower socioeconomic status alongside protective factors such as coping skills and strong social support. A third model synthesizes those assessments into a structured summary, and a final layer combines that summary with demographic information, trauma type and known PTSD prevalence rates to produce a risk score with a plain-language explanation.

Experiments found that the tool’s three-layer design consistently outperformed simpler configurations across most performance metrics. The researchers also compared its results with diagnoses for the same cases using the Clinician-Administered PTSD Scale for DSM-5, or CAPS-5, the gold standard clinical interview for diagnosing PTSD.

Emergency departments rarely screen routinely

About 43 million patients are admitted to U.S. emergency departments each year with physical injuries, and more than a quarter of them develop PTSD, according to past studies. Only about 28 percent of emergency departments routinely screen for PTSD, and only two-thirds of those use validated tools.

The study authors say the gap likely reflects overcrowding and limited clinician time: clinicians typically have about 15–20 minutes per patient and must prioritize immediate medical concerns over lengthy mental-health assessments. The researchers describe the first month after trauma as a key period for identifying people at risk, because many patients are no longer in contact with the hospital after that.

Schultebraucks said PREVIA uses open-source models and requires no training or fine-tuning. The study was funded by National Institute for Mental Health grant R01 MH129856, with Schultebraucks as principal investigator.