Review of Machine Learning Algorithms for Autism Spectrum Disorder Prediction

Autism is a disorder of the brain caused by language and severe social difficulties in people. It is the most common of the many pervasive developmental disorders (PDD), which begin in childhood and continue throughout life, affecting almost all aspects of life. Autistic people (cognitive and linguistic) and social skills are delayed compared to their classmates, but their motor skills are higher than usual. The nature of ASD varies from person to person and is a condition for the development of the brain or nervous system. In the first 12 months, most children with autism seem to develop normally. Symptoms of autism appear between eighteen and thirty-six months. Up to 3 years, 40% of cases were detected. The aim of the study was to detect ASD at an early stage to improve brain development and increase the awareness of parents and caregivers about ASD. Machine learning methods are now used to predict the spectrum of autism. This study provides a comprehensive assessment of documents that use machine learning to predict ASD, as well as data analysis and classification algorithms. This work aims to classify and study the different methods of Machine Learning, as well as to explain the nature of ASD and to evaluate performance and demonstrate research potential using different criteria. This publication serves as a roadmap for imminent researchers who want to work on the topic of ASD prediction using machine learning.