Individualized <scp>EEG</scp> functional connectivity predicts clinical symptoms in <scp>ADHD</scp> , dyslexia, and their comorbidity

BACKGROUND: Attention-deficit/hyperactivity disorder (ADHD) and developmental dyslexia (DD) are two prevalent neurodevelopmental disorders (NDDs) that frequently co-occur, resulting in comorbidity (COM), which is associated with significant cognitive and behavioral difficulties. These disorders are characterized by considerable clinical and neurobiological heterogeneity. The functional connectivity (FC) patterns within each disorder may consist of both individual-shared and individual-specific components. Disentangling these patterns may provide more in-depth insights into the neural mechanisms underlying these conditions. METHODS: In this study, the common orthogonal basis extraction (COBE) method was applied to the resting state electroencephalography (EEG) FC data from children with ADHD, DD, COM, and age-matched typical development (TD) controls, aiming to separate individual-shared and individual-specific FC patterns. RESULTS: Individual-specific FC could significantly predict clinical symptoms in ADHD, DD, and COM, with distinct predictive connectivity patterns across different groups. The predictive features in ADHD were primarily located in the left central and parietal regions. In DD, they involved the left temporal lobe, left parietal lobe, and right central region. In COM, they involved the left frontal lobe and the right parietal lobe. At the group level, compared with TD, individual-shared FC showed significantly enhanced posterior FC in all clinical groups. The differences in shared FC among the three clinical groups further allowed for the classification of ADHD, DD, and COM with an accuracy of 73.55%. CONCLUSIONS: By disentangling the shared and individualized components of EEG FC, this study provides novel insights into the neural organization of ADHD, DD, and their COM. The results suggest that individual-specific FC is more strongly associated with clinical symptoms, whereas shared FC captures group-level connectivity alterations and differences among clinical groups. This framework may present a promising approach for investigating biological heterogeneity in NDDs.