Autism spectrum disorder (ASD) is characterized by clinical and etiological heterogeneity. Although there is broad agreement on the necessity of integrating reliable biomarkers of ASD, little is known about how epigenetic and/or brain factors contribute to ASD. To probe epigenetic factors, we obtained the participants’ saliva samples to obtain DNA methylation values of the oxytocin receptor and arginine vasopressin genes. Given the atypical sensory processing associated with ASD and the role of thalamus as the sensory relay station, we acquired neuroimaging data to create a thalamocortical resting state functional connectivity map of ASD. Integrating features derived from epigenetic, brain, and sensory behavior factors, the current study aims to develop a model that predicts ASD using an eXtreme Gradient Boosting algorithm with ensemble feature selection procedure based on feature occurrence frequency and automatically optimized hyperparameters. We found the full model with all three factors (83.95%) to be superior for ASD classification, compared to models with epigenetic and behavioral factors (76.89%) or with brain and behavioral factors (81.19%).