An improved adaptive neuro fuzzy interference system for the detection of autism spectrum disorder

Abstract

Autism spectrum disorder (ASD) has become one of the most areas of research in present decade. Previous research works have focused on varied approaches to handle Autism spectrum disorder. However, significance of the existing approaches could still be improved. The efficiency and accuracy of ASD prediction can be improved via building classification systems using artificial intelligence mechanisms like machine learning. This paper addresses this issue by designing an expert system to evaluate autism under uncertainty. An improved adaptive neuro fuzzy interference system (IANFIS) approach has been employed to develop an autism detection model for detecting ASD in individuals of all ages. The advantage of IANFIS is to give the highest classification and the lowest error rates when compared to other existing classifiers. Here the Particle Swarm Optimization (PSO) method is proposed for identifying and choosing the most important ASD feature subset. Its performance was evaluated with ISAA dataset, which comprises of data from individuals with and without autistic attributes. Experimental outcomes revealed that the proposed ASD detection model generates better outcomes in terms of sensitivity, accuracy, precision and specificity.