Data Mining for Autism Spectrum Disorder detection among Adults

Autism Spectrum Disorder (ASD) is one of the most common children's neurodevelopmental disorders (NDD) with an estimated global incidence of 1% to 2%. There are two aims for this research, first, to propose a data mining architecture that combines behavioural and clinical characteristics with demographic data. Second, to provide a quick, acceptable and easy way to support the ASD diagnosis. this can be performed by conducting a comparison study to determine the efficacy of four possible classifiers: logistic regression (LR), sequential minimum optimization (SMO), naïve Bayes, and instance-based technique based on k-neighbors (IBK). These classifiers have been performed with Waikato Environment for Knowledge Analysis (WEKA) tools to distinguish autistic adults from healthy, normal subjects. The results showed that, with 99.71%, SMO classification accuracy was 99.71, which exceeded the accuracy of other classifiers. The proposed architecture allows for early detection of ASD, distinguishing between ASD and healthy control subjects. This study could help doctors and clinicians by giving them a better idea of what the future holds for people with autism spectrum disorder (ASD) and by improving therapy programs, allowing people with ASD to live a long and happy life.