Innovative study to develop better diagnostic tools for PTSD

$2.3 million CDC Grant to develop PTSD Symptom Tracer and Outcome Prognosticator

Posttraumatic stress disorder (PTSD) is common in the general population, and many responders to World Trade Center (WTC) attacks continue to suffer from this disorder.

Efforts to reduce PTSD are hindered by the limitations of existing diagnostic tools. Diagnostic interviews and self-report scales are subjective, which reduces precision and accuracy of diagnosis. Currently, there are no objective measures of PTSD available for clinical use. As the result, diagnosis provides limited guidance in selecting psychological treatment for individuals who have suffered trauma.

The study team applies Artificial Intelligence (AI) to examine audio, video, and daily activity for patterns associated with PTSD over time. This approach will produce PTSD-STOP (PTSD Symptom Tracer and Outcome Prognosticator), an AI model with the capabilities of:

  1. monitoring symptom severity objectively and
  2. forecasting symptom exacerbation and treatment response. 

Participants in this study are patients with PTSD symptoms treated at the Stony Brook WTC Health Program and other clinics. Participants are invited to make brief video recordings and complete mini-surveys daily over a 3-month period. This will allow investigators to develop AI that tracks PTSD symptoms across language, speech, facial expression, and activity. Additionally, an AI is created to predict future symptoms from past patterns of behavior measured objectively. Clinical usefulness of PTSD-STOP will be tested by comparing it to traditional diagnostic methods in predicting clinical improvement, service utilization, severity of traumatization, and genetic risk.

This proposed study stands as the first initiative to utilize AI for monitoring PTSD symptoms over time. Overall, we propose to use innovative technologies to address problems that limit care of patients with PTSD. While the initial version of PTSD-STOP will serve as a research tool, it holds potential for further development into clinical applications.  Ultimately, it will be possible to use PTSD-STOP with unobtrusively collected data (e.g., recordings of telehealth visits).

This study is a collaborative effort between the departments of Computer Science and Psychiatry, as well as the WTC Health Program. The investigators include Hansen Andrew Schwartz, PhD (Computer Science), Roman Kotov, PhD (Psychiatry), Dimitris Samaras (Computer Science), and Benjamin J. Luft, MD (WTC Health Program).