Rhonda Brandon Senior Vice President and Chief Human Resources Officer, DUHS | Duke University Hospital
Rhonda Brandon Senior Vice President and Chief Human Resources Officer, DUHS | Duke University Hospital
An AI model developed by Duke Health researchers has shown promise in predicting when adolescents may be at high risk for serious mental health issues. This model, which goes beyond identifying existing symptoms, pinpoints underlying causes such as sleep disturbances and family conflict. Such capabilities could potentially widen access to mental health services through primary care providers.
Jonathan Posner, M.D., a professor in the Department of Psychiatry and Behavioral Sciences at Duke, emphasized the urgency of addressing youth mental health. "The U.S. is facing a youth mental health crisis -- nearly half of teens will experience a mental illness," he stated. He further noted the shortage of mental health providers in the U.S., suggesting that this AI model could empower pediatricians to intervene early.
The research team, including data scientists Elliot Hill and Matthew Engelhard from Duke’s Department of Biostatistics & Bioinformatics, utilized data from the ABCD study involving over 11,000 children. They developed a neural network to predict psychiatric risk escalation within a year with an accuracy rate of 84%.
Additionally, an alternative model was tested to identify potential mechanisms leading to worsening mental illness, achieving an accuracy rate of 75%. The identification of underlying causes offers unique insights for potential interventions.
Elliot Hill explained the significance: “It’s important to leverage that information to design an intervention for that child.” Among identified risk factors, sleep disturbances were highlighted as a strong predictor.
Matthew Engelhard clarified that while sleep disturbances are not proven causes, they are associated with high-risk rates. The authors suggest this tool could help primary care practitioners assess children's mental health risks using simple questionnaires.
Posner added that time constraints often prevent detailed psychiatric assessments in primary care settings. This AI model would automate data analysis and provide doctors with clear indications of risk levels.
The study's authors include Pratik Kashyap, Elizabeth Raffanello, Yun Wang, Terrie E. Moffitt, and Avshalom Caspi. Funding support came from various institutes including the National Institute of Mental Health and the National Institute on Aging.
###
Alerts Sign-up