Stanford researchers use speech patterns to predict childhood mental health risks

Researchers at Stanford University have developed a method using natural language processing (NLP) to predict the risk of mental health disorders in children. According to a study published in Nature Mental Health, analyzing how children speak about stressful life events can predict the development of mental health issues six years later with higher accuracy than a panel of experts.

The study found that the linguistic structure—specifically how children use conjunctions and short auxiliary words like ‘and’, ‘to’, or ‘but’—was a more significant predictor than the actual content of their stories. Lead author Chase Antonacci noted that this could provide a scalable tool for identifying risks before clinical diagnoses are made, which is critical as depression and anxiety often emerge during adolescence.

Co-author Ian Gotlib highlighted that speech analysis is a cost-effective and accessible alternative to biological markers like cortisol levels or telomere length, potentially offering a superior indicator for future mental health developments.