Fig. 3. Machine-learning (ML) approaches for early diagnosis and better evaluation of autism-spectrum disorder (ASD).

ML methods or techniques can be applied to ASD raw data obtained from genomics and neuroimaging approaches. ML uses supervised and unsupervised learning methods to classify and distinguish clusters in ASD. Deep-learning techniques, such as deep neural network (DNN) and convolutional neural network (CNN), can be applied to diagnose ASD.