Data-Driven Identification of Brain–Behavioral and Sociodemographic Predictors of Anxiety Severity in Children Using Machine Learning

This is the authors' abstract. We don't add key points for this paper.

Objective: Childhood anxiety is highly prevalent and linked to cognitive control deficits and sociodemographic risk factors. However, how these factors contribute and interact remain unclear. This study used a data-driven machine learning approach to identify the strongest neural, behavioral, and sociodemographic predictors of childhood anxiety severity. Method: -Oriented Anxiety Scale. Predictors included cognitive control markers derived from electroencephalogram recordings and behavioral measures, both obtained during a Go/NoGo task, along with sociodemographic risk factors. Random Forest, Support Vector Regression, and XGBoost models were compared using cross-validation. Partial dependence plots were used to examine predictor-outcome relationships, and interactions were analyzed using H-statistic. Results: Random Forest yielded optimal prediction accuracy. Error positivity (Pe) emerged as the strongest predictor, followed by single-mother status, frontal theta power, post-error accuracy, and occipital post-error alpha power. Single-mother status strengthened the relationship between post-error alpha power and anxiety, indicating that impaired post-error attentional mechanisms may be more pronounced in anxious children from single-mother households. Low frontal theta power amplified the Pe-anxiety relationship, suggesting that reduced inhibitory control may intensify anxiety-related error awareness. Conclusion: A specific interaction among neural indices of cognitive control (Pe, frontal theta, occipital post-error alpha), behavioral performance (post-error accuracy), and a sociodemographic factor (single/unpartnered mother) predicted childhood anxiety. These findings highlight the importance of considering both neural mechanisms and environmental context in guiding diagnosis and treatment, offering insights beyond traditional hypothesis-driven approaches. Diversity & Inclusion Statement: We worked to ensure sex and gender balance in the recruitment of human participants. We worked to ensure race, ethnic, and/or other types of diversity in the recruitment of human participants. We worked to ensure that the study questionnaires were prepared in an inclusive way. One or more of the authors of this paper self-identifies as a member of one or more historically underrepresented racial and/or ethnic groups in science. One or more of the authors of this paper self-identifies as living with a disability. One or more of the authors of this paper self-identifies as a member of one or more historically underrepresented sexual and/or gender groups in science. One or more of the authors of this paper received support from a program designed to increase minority representation in science.