Machine Learning Methods to Detect Autism Among Children: Review

Autism Spectrum Disorder (ASD) is a neurological condition that permanently affects a person’s ability to learn languages, speak, think clearly, and interact with others. In the embryonic stages, its symptoms typically emerge. The conditions of ASD can be improved by identifying and treating them early. ASD is either genetic or caused by environmental factors. Currently, the only methods utilized to diagnose ASD are clinically based-tests. In addition to requiring extended diagnostic time, this will also result in a sharp increase in medical expenses. Machine Learning (ML) approaches are being used to supplement traditional methods to increase diagnoses’ accuracy and response time. The paper presents a review of the recent literature on different ML approaches for the classification and identification of ASD in children and the study of the characteristics of ASD. The concept of ASD and the techniques used to discover it are explained; at the end of the paper. The most prominent techniques used in ML to identify children with ASD were discussed in detail. Finally, The main objective of our study is to focus on whether the child is at risk of developing ASD spectrum disorder in its early stages And highlight the importance of early diagnosis using ML methods to reduce the symptoms of the disease.