An AI-based tool called PREVIA can accurately identify trauma patients who will go on to develop posttraumatic stress disorder (PTSD), according to a report published today in Nature Communications. PREVIA analyzes patient narratives about how they ended up in the emergency department (ED).Why It’s Relevant
While there are screening tools available to identify injured patients who are at risk for PTSD, they’re not often used in the ED given limited clinician time (screeners take about 10-15 minutes) and concerns that initiating mental health conversations will increase patient discomfort. An automated tool that can assess patterns in a patient’s response to a standard intake question such as “What brought you to the ED?” could be an answer to this screening gap.
By the Numbers
- Researchers tested PREVIA on 145 patients (69% male) across two New York City trauma centers. Patient narratives were recorded within 24 hours of injury, and clinicians tested for PTSD one month later using the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5).
- PREVIA identified 41 of the 43 patients who were subsequently diagnosed with PTSD and had an overall accuracy of discriminating PTSD versus no-PTSD of 78.5%.
- In a head-to-head comparison of 30 random patients from the main group, PREVIA had comparable accuracy to an ED experienced clinician who evaluated the narratives.
What’s More
PREVIA remained accurate when the CAPS-5 cutoff score was lowered to include individuals who developed subthreshold PTSD—a category often missed by traditional screeners but that still carries elevated health risks—correctly identifying 89.7% of positive cases (52 of 58).
The Other Side
Though PREVIA had an exceptional rate of identifying at-risk patients, the program also flagged many false positives, leading to a modest overall precision (true positives + false positives) of 50%. This means that, in practice, using PREVIA would still require time and resources for follow-up evaluations. Further, the researchers excluded severe trauma patients who were unable to provide consent, potentially omitting a group with the highest risk of future PTSD.
Takeaway Message
“We don’t want to replace clinicians, nor do we think it’s even possible,” senior study author Katharina Schultebraucks, Ph.D., told Psychiatric News. But an automated tool like PREVIA can be a valuable assistant, she continued, identifying potential future events while ED staff focus on the present. “We can flag at-risk patients and follow up with them, and perhaps be in a position to intervene before they require intensive treatment."
Related Information
Source
Victoria Mueller, et al. LLM ensemble PREVIA predicts future PTSD risk from emergency department trauma narratives with expert-level accuracy. Nature Communications. Published November 9, 2026. doi:10.1038/s4167-026-78768-3
(Image: Getty Images/iStock/JazzIRT)
