Deep Learning Model for Predicting Alcohol Use Disorder Risk in Adolescents and Young Adults

Researchers developed a deep learning model using longitudinal data to predict individualized risk of developing alcohol use disorder in adolescents and young adults who drink, identifying biological sex, delinquency, and personality traits as key risk factors. The model demonstrated good predictive performance (AUC 0.86 at 1-6 year follow-up) with appropriate validation and calibration metrics.