Information-based Multivariate Decoding Reveals Imprecise Neural Encoding in Children with Attention Deficit Hyperactivity Disorder During Visual Selective Attention.

Attention deficit hyperactivity disorder (ADHD) is a common neurodevelopmental disorder in school-age children. Attentional orientation is a potential clinical diagnostic marker to aid in the early diagnosis of ADHD. However, the underlying pathophysiological substrates of impaired attentional orienting in childhood ADHD remain unclear. Electroencephalography (EEG) was measured in 135 school-age children (70 with childhood ADHD and 65 matched typically developing children) to directly investigate target localization during spatial selective attention through univariate ERP analysis and information-based multivariate pattern machine learning analysis. Compared with children with typical development, a smaller N2pc was found in the ADHD group through univariate ERP analysis. Children with ADHD showed a lower parieto-occipital multivariate decoding accuracy approximately 240-340 ms after visual search onset, which predicts a slower reaction time and larger standard deviation of reaction time. Furthermore, a significant correlation was found between N2pc and decoding accuracy in typically developing children but not in children with ADHD. These observations reveal that impaired attentional orienting in ADHD may be due to inefficient neural encoding responses. By using a personalized information-based multivariate machine learning approach, we have advanced the understanding of cognitive deficits in neurodevelopmental disorders. Our study provides potential research directions for the early diagnosis and optimization of personalized intervention in children with ADHD.