The Added Value of Neuro-informed Assessment of Externalizing Behavior Problems: A Prospective Population-Based Neuroimaging Study

Externalizing behavior problems in youth are associated with substantial economic, emotional, and health-related consequences for youths, their families, and society. While prior research has identified various neurobiological markers linked to externalizing behavior, the predictive value of brain network metrics for externalizing problems in young people remain unexplored. This study aimed to identify predictors of externalizing behavior problems in children using the longitudinal Generation R Study. Functional MRI (fMRI) data were acquired at age 10 and 14, and reconstructed into network metrics. These metrics were analyzed alongside behavioral and environmental factors, including the syndrome scales of the Child Behavior Checklist (CBCL), callous traits, demographic characteristics and stressful life events at age 10, to develop a random forest model predicting externalizing behavior problems at age 14 years (N = 640). The best-performing model achieved an accuracy of 78.0%, with externalizing behavior at age 10 emerging as the most important predictor (variable importance of more than 60%). Additional variance explained by including neurobiological variables was minimal (only 1%). Early screening of externalizing behavior at a younger age remains a promising and cost-effective avenue for identifying at-risk children. Future research should further explore the added value of other neurodevelopmental metrics (e.g. task-based fMRI, EEG, genetics) in neuro-informed assessment studies of externalizing problems in youth – that also encompasses non-brain metric comprehensiveness – and explore how prediction-based outcomes can guide timely diagnosis and treatment to mitigate the long-term burden of externalizing behavior problems.