EEG Feature Framework Predicts Psychopathology Dimensions.

Researchers have introduced a novel framework for analyzing electroencephalography (EEG) data to predict dimensional psychopathology. The framework, termed a granularity-aware EEG feature pipeline, systematically organizes multi-scale descriptors into global, regional, and channel levels, allowing for a comprehensive examination of neurophysiological correlates. The study utilized the Healthy Brain Network (HBN) cohort to evaluate the prediction of four key psychopathology dimensions: p-factor, internalizing, externalizing, and attention problems, across four distinct EEG paradigms. Given the inherent heterogeneity of pediatric psychopathology and the moderate reliability of questionnaire-derived scores, this investigation served as a challenging feasibility test rather than a direct clinical screening tool. The findings indicate that while effect sizes remained modest, tree-based models combined with granularity-balanced feature selection demonstrated promising improvements compared to traditional methods in specific conditions. Visualizations of the selected markers revealed spatial and spectral patterns consistent with existing neurophysiological knowledge, and an exploratory cross-dataset check suggested the selection principle's technical feasibility across different protocols.