Atypical Dynamics of Sub-Networks Predict Restricted Repetitive Patterns of Behaviors in Children with Autism Spectrum Disorder

Abstract

Previous studies indicated that the atypical dynamics may underlie the restricted, repetitive patterns of behaviors (RRB) in Autism spectrum disorder. However, the temporal architecture of ASD remains unclear. Here, we developed matrix factorization method to decompose the dynamic functional network into sub-networks and weights (which embed the temporal features of sub-networks) and applied this model to a large sample size and multi-site resting-state functional magnetic resonance imaging data of 105 children with ASD and 102 matched typically developing controls, which acquired from the Autism Brain Imaging Data Exchange dataset. Compared to TDC, the sub-networks exhibited atypical average and variance of weights in ASD. Moreover, these temporal features can predict RRB scores. Overall, our studies provided a subnetworks-based perspective to explore the atypical temporal features and relationship between these temporal features and RRB symptom.