A Data Mining Based Approach to Predict Autism Spectrum Disorder Considering Behavioral Attributes

Abstract

Autism Spectrum Disorder (ASD) is a condition that hinders brain development. It affects a person’s way of communicating and behaving. It impacts how a person’s way of perceiving and socializing with other people. People with ASD experience different type of symptoms like difficulty with interacting with others, repetitive behaviors, difficulty to function properly in all areas of life. And these symptoms generally occur in early childhood. In this paper, a data mining classification technique was used for the prediction of ASD in adults. Random forest was used as the classifier and accuracy, sensitivity and specificity score 0.9946, 0.9874, 0.9975 were obtained from training set and 0.9571, 0.8571, 0.9821 were obtained from testing set. The dataset used for prediction had 10 behavioral attributes and 10 more individual attributes.