USF Develops AI Tool To Detect PTSD in Kids Through Facial Cues

A new AI tool from USF detects PTSD in children by analyzing facial expressions, offering clinicians a private, non-verbal way to assess trauma.

The researchers of this new study explain that diagnosis of PTSD in children is extremely challenging because children find it difficult to communicate their traumatic experiences through words.

To tackle this issue, the researchers at the University of South Florida (USF) have developed an AI-powered diagnostic tool, which shows promise to transform current practices. The system uses facial expression analysis to identify PTSD symptoms in children, thus providing clinicians with an objective, cost-effective trauma assessment method.

Led by USF professors Alison Salloum, an expert in childhood trauma, and Shaun Canavan, a specialist in AI and facial analysis, the project combines real-time emotion recognition with patient privacy protection. “Avoidance is a main component of PTSD, so children don’t want to talk about it,” Salloum said to Tampa Bay Times (TBT). “There are lots of reasons — one is just the sheer horror of what happened.”

Traditional methods rely on interviews and questionnaires, which often fall short because children lack the cognitive ability to express their feelings, or fear disturbing their family members.

However, Salloum observed that children displayed hidden emotions through their facial expressions during virtual interviews. She approached Canavan about the possibility of technological detection of these moments. “I was confident it would work,” he said to TBT.

The research team worked with anonymized video data from 18 children to record hundreds of thousands of video frames. The AI system detected the minimal emotional responses which are typical for PTSD patients. The researchers made sure to store no raw video content.Only facial movement data and contextual information were used.

The study is the first to incorporate context-specific PTSD classification while preserving user privacy. The team hopes to develop a real-time system, which will include a clinician interface.

Salloum emphasized that AI won’t replace human evaluation but will support it. “It doesn’t replace the other methods of asking questions, conversations and interviews about what the person has experienced,” she said to TBT. “It’s really a tool,” she added.

The researchers say to TBT that they hope that the system may one day help other groups too, like combat veterans or abuse survivors, offering clinicians a new lens into trauma.