Advances in the diagnosis of autism spectrum disorder through machine learning.

Autism spectrum disorder (ASD) is a neurodevelopmental condition that is currently diagnosed by behavioural assessment. No biomarker is used in its diagnosis due to the poor understanding of its pathogenesis. Distinct features have been discovered in magnetic resonance imaging (MRI) of brains with and without ASD. Promising diagnostic results have been achieved using features constructed from various MRI modalities. This review explores the extent to which ASD can be diagnosed through machine learning applied to MRI brain data. Several exciting biomarkers have also been proposed, which identify physiological alterations that may be diagnostic for ASD.