Emory researcher uses artificial intelligence to predict youth mental health risks

Caption

A study of more than 11,000 children aged 9 to 10 found that sleep disturbances were the strongest predictor of future psychiatric illness.

Credit: Pexels

Sleep problems in your tween might mean it's time for a mental wellness check, according to a new study published in Nature Medicine.

Yun Wang, a neuroscientist and assistant professor at Emory University School of Medicine who co-authored the study, developed a model that predicts future psychiatric risk using artificial intelligence.

Sleep disturbances were the strongest predictor of future psychiatric illness, surpassing childhood trauma and family history, she said.

Caption

Yun Wang, Ph.D, assistant professor at Emory University School of Medicine.

Credit: Emory University School of Medicine

They examined data from the ongoing Adolescent Brain and Cognitive Development (ABCD) Study, which includes over 11,000 children who participated in multiple assessments of their psychosocial environment and brain development collected over five years.

It's the largest study of its kind using AI to track and predict youth mental health risks over time, she said.

Surprisingly, adding neuroimaging data did not improve the AI’s predictive accuracy, making this a highly scalable and accessible screening tool.  

"It's good news because MRI scans are expensive," Wang said. "We can use low-risk and low-cost questionnaires to predict [any] future mental health issues."

By studying the underlying causes of mental illness — as opposed to focusing only on symptoms of disease — researchers intended to identify risk and guide prevention strategies.

The next step, Wang said, is to try to replicate these results in a clinical setting. 

If sleep is still the strongest predictor of future mental illness, then clinicians could develop interventions for those identified as at high risk.