EEG Microstates and Machine Learning for Objective ADHD Diagnosis in Children and Adolescents

This study proposes a machine learning model using resting-state EEG features to objectively identify ADHD and predict symptom severity in medicated and unmedicated children and adolescents. The model achieved strong training performance (AUC 0.967) but more modest validation results (AUC 0.747), suggesting the approach shows promise but requires further independent validation before clinical implementation.