AI Model Predicts ADHD Using VR and Eye Data
A new study in Translational Psychiatry shows that combining virtual reality (VR) with eye tracking, head movement data, and self-reported symptoms can improve the accuracy of diagnosing ADHD in adults. Researchers developed a machine learning model that correctly identified ADHD in 81% of cases using data from a VR-based attention task in a realistic, distracting environment.
The study involved 86 participants and showed that features like gaze wandering, variable reaction times, head movement, and symptom self-reports were the most useful for diagnosis—while EEG data did not enhance accuracy. This multi-method approach, validated on a separate test group, offers a promising path toward more objective and ecologically valid ADHD diagnostics, especially given current reliance on subjective interviews.
Researchers emphasize the need for larger studies and aim to create a standardized diagnostic tool for clinical use.