Autism spectrum Disorder Prediction with Binary Value and EEG signal by Implementing Decision tree and ELM algorithm

Abstract

Autism disorder is a developmental disorder that affects communications and behaviour of a child or a person. The symptoms for this disorder appear in the first two years of life. This disorder is not hard to detect, but it needs persons with proper training and experience. There are several reasons for this disorder and some of the main cause are difference in the brain, genetic and environmental conditions. This also includes limited and repeated patterns of behavior. Detecting autism through screening tests is very expensive and time consuming. To avoid this complexity this research paper proposed a new work with the combination of machine learning and deep learning algorithm. This integration algorithm provides stable and accurate decision in predicting autism disorder. Decision tree algorithm to predict the person (or) a child affected with this disorder or not and to improve the accuracy Extreme learning machines (ELM) is used. All the results are compared with KNN, ANN. Doing predictions using EEG signals the accuracy got improved and the results are compared with normal binary value dataset. To test this algorithms Kaggle dataset is used for training, testing for the prediction of this disorder and to improve the decision making.

Keywords: Autism Spectrum Disorder, Extreme learning machines, Decision tree algorithm, Kaggle ASD.