Detecting Autism Spectrum Disorder Using Data Mining

Abstract

AbstractAutism spectrum disorder (ASD) is a set of neurodevelopmental disorders that affect cognitive development, social and communication skills, and behavior of affected individuals. The faster traces of ASD are identified, the faster the stimulation will begin and the more effective the gains in neuropsychomotor development will be. That being said, the earlier the diagnosis of ASD, the easier it is to control the disorder. Therefore, this study aims to classify the cases of ASD as “yes” if a patient has been diagnosed with ASD or “no” if a patient has not, using data mining (DM) models with classification techniques. The methodology of cross-industry standard process for data mining (CRISP-DM) was followed, and to induce the data mining models, the Rapidminer software was used. The results were quite promising reaching a level of accuracy of 97%, specificity of 95.45%, sensitivity of 100%, and precision of 95.65%.KeywordsAutism spectrum disorderData miningCRISP-DMClassification