Prognostication of autism spectrum disorder (ASD) using supervised machine learning models

Abstract

Autism spectrum disorder (ASD) screening is a psychiatric disorder which leads to neurological and developmental growth of a person which begins early in childhood and lasts throughout a person’s life. This disorder is caused by differences in the brain, genetics and environmental conditions. The disorder also includes limited and repetitive patterns of behaviour. It affects how a person interacts with others, communicates, and learns. The main areas of functioning affected in people with ASD qualitative impairments in social interaction and qualitative impairments in communication. Among the affected, it was observed that more men were affected with this disorder when compared to women. The cases related to these disorders are increasing progressively. The centres of disease control and prevention (CDC) currently estimate that one in 59 children is diagnosed with ASD disorder. Unfortunately, waiting time to diagnosis these disorders are lengthy and procedures are not cost effective. To overcome the time complexity for identifying the disorder, computational intelligence can be used by making use of advanced technologies such as machine learning to improve precision, accuracy and quality of the diagnosis process. Machine learning helps us by providing intelligent techniques to discover useful hidden (or) concealed patterns, which can be utilized in prediction and to improve decision making.