Precision Diagnosis of ADHD Subtypes with Integrated Multimodal Data

Attention deficit hyperactivity disorder (ADHD) is one of the most prevalent neurodevelopmental disorders impacting children and adolescents. In neurodevelopmental research, ADHD subtypes provide a major issue since there are not enough reliable biomarkers to classify them. This study investigates the predictive potential of ADHD subtypes using a multimodal dataset, including behavioural measures, self-reported symptoms, and EEG data. The study utilized Principal Component Analysis (PCA) to reduce dimensionality, while Random Forest (RF) classifiers were employed for predicting subtypes. The research work achieved a cross-validation accuracy of 96.6% and test accuracy of 95.2%, outperforming both KNN and SVM. Feature importance analysis highlighted that self-reported symptoms were the most significant predictors of DSM subtypes, with neuropsychological data outperforming EEG features for subtype classification. These results demonstrate the effectiveness of several measurement modalities in categorizing ADHD subtypes and emphasize the necessity of accurate assessment methods to improve diagnostic precision and guide focused intervention practices.