Using Machine Learning To Identify Neural Mechanisms Underlying the Development of Cognition in Children and Adolescents With ADHD
Abstract
Childhood and adolescence are marked by improvements to cognition and by the emergence of neurodevelopmental disorders such as attention deficit hyperactivity disorder (ADHD). What neural mechanisms are associated with cognitive development in ADHD? In this study, I applied machine learning models to functional connectivity profiles to identify patterns of network connectivity that predict various cognitive abilities in a group of participants ages 6 to 16 with ADHD. The models successfully predicted IQ, visual spatial, verbal comprehension, and fluid reasoning in children ages 6 to 11, but not adolescents. Furthermore, the models identified connections with the default mode, memory retrieval, and dorsal attention networks as driving prediction during early and middle childhood, but connections with the somatomotor, cingulo-opercular and frontoparietal networks were more important in middle childhood. These results suggest that computational models can identify neural mechanisms associated with predicting cognitive abilities in children and adolescents with ADHD using naturalistic stimuli.