This study reveals alterations in alpha-band functional connectivity in autistic children. We observed weaker connectivity in the frontal lobe in individuals with autism. Based on this finding, we suggest placing greater emphasis on sampling EEG data from the frontal lobe. Additionally, this study highlights the critical importance of careful selection and interpretation of input features when constructing classifiers. Moreover, the obtained results demonstrate that contrastive variational autoencoders are a promising approach for discovering latent patterns and features in complex networks.
(DOI: 10.1016/j.chaos.2024.115123)