Application and evaluation of classification model to detect autistic spectrum disorders in children

Abstract

An Autism Spectrum Disorder (ASD) affected child faces significant difficulties in social interactive activities. There is an eminent requirement of a real-time and easy-to-access diagnostic model to identify associated risks during initial phase of occurrence for proper diagnosis. This research deals with efficient categorisation of ASD instances using an interactive and intelligent classification model where a child can be detected with autism disorders through an automated interactive queries enabled virtual session. Implementation outcome indicates that random forest method generated an optimal performance recording an accuracy rate of 97.5% and an error rate of 0.676. Also a least latency of 1.16 sec was noted. 92.8%, 92.6%, 90.8% and 91.5% are the generated mean accuracy, precision, recall and F-score metrics respectively. Thus, it can be concluded that autism risks detection with random forest classifier can assist medical experts in accurate diagnosis.