A team of researchers at the University of South Florida has broken new positions in the diagnosis and treatment of post-traumatic stress disorder (PTSD) in children. Alison Salm, a professor at the Faculty of Social Work, and Sean Canavan, an associate professor at Bellini University for artificial intelligence, cybersecurity and computing, has developed a pioneering artificial intelligence system that analyzes facial expressions to help clinicians identify PTSD in young patients.
Diagnosis of PTSD in children is usually based on subjective methods such as clinical interviews and self-reported questionnaires, both of which may be limited by the child’s stage of development, language skills, or emotional avoidance. Salloum’s clinical experience sparked the idea of a more objective solution. “Some children’s expressions became very intense during the trauma interview,” Saloom said. “Even when they didn’t put much into words, their faces made such a clearer. That’s when I asked Sean if AI would help us analyze it.”
Canavan is in facial analysis and emotional recognition, Canavan adapts tools from the lab and creates a privacy-first system that processes unidentified video data. “We don’t use any live videos at all,” emphasized Canavan. “AI only analyzes facial movement data and considers it in the context of whether the child is talking to a clinician or parent.”
Peer-reviewed studies from the team published in Science Direct first used context-aware PTSD classification techniques while retaining full anonymity of participants. Using data from 18 treatment sessions with children, the researchers analyzed more than 100 minutes of footage per participant, and analysed a total of hundreds of thousands of video frames. The AI system identified subtle yet consistent facial muscle movements associated with PTSD, particularly during interviews with clinicians. This became more clear than the parent-child conversation.
“This is not a replacement for a clinician,” explained Saloom. “It’s about giving them an extra layer of insight. We can imagine this system as a real-time tool during treatment sessions and help us monitor the child’s progress without repeated and potentially traumatic evaluations.”
The research team plans to expand its research to examine potential biases across gender, culture and age. Particularly among preschool children with limited oral communication and diagnosis often relies on parental observation.
While still in its early stages, the project already demonstrates its promise. In particular, we consider the complex clinical profiles of many participants who presented joint conditions such as depression, ADHD, and anxiety.
“This high quality data is rare in AI,” Canavan said. “We are proud to have designed research that is not only innovative but ethically sound. This tool could ultimately help clinicians to provide better information and improve mental health care for some of the most vulnerable populations.”
When validated in large trials, the USF team’s system could mark a transformative change in how PTSD is diagnosed and tracked in children. This shapes AI and video technology to bring greater accuracy and empathy to pediatric mental health care.
