An NIH-funded study found that how young people linguistically frame stressful events may predict the onset of depression and anxiety disorders years before symptoms emerge. The findings, published in late July, offer a potential tool for early intervention in the nation’s youth mental health crisis.
Researchers used AI-based natural language analysis tools to identify patterns in children’s speech associated with future mental health disorders. The predictive markers could enable clinical intervention before a child develops diagnosable symptoms.
Mental health disorders among American youth have reached crisis levels. The U.S. Surgeon General declared a youth mental health emergency in 2021, and rates of depression, anxiety and self-harm among children and adolescents have continued to rise.
The study found specific linguistic patterns — including how children describe the causes, duration and controllability of negative events — that correlated with later development of depressive and anxiety symptoms. Children who used more global, stable and internal attributions for negative events were at higher risk.
The AI-based analysis approach is potentially scalable in ways that traditional clinical screening is not. If validated in larger populations, language-based screening tools could be deployed in school settings, pediatric offices or even through digital applications, reaching children who might never receive a formal mental health evaluation.
The research adds to a growing body of work using machine learning to extract clinically actionable insights from language and behavior patterns. Previous studies have explored similar approaches for predicting psychosis, suicidal ideation and cognitive decline.
The NIH’s National Institute of Mental Health funded the research as part of its broader investment in early identification and prevention of mental health disorders.