Chaitra Bhaskarachary, Amir Jahanian Najafabadi, Ben Godde

• Autism Spectrum Disorder is a neurological and developmental disorder that starts early in adolescence and lasts throughout a person’s life affecting information flow in the brain leading to secondary problems for the patient. Early detection of ASD is vital in enhancing the efficiency of the treatment. Current diagnostic approaches for autism are time-consuming, to accelerate this process of diagnosing the disease as early as possible with fewer efforts and better accuracy machine learning methods have been proposed recently. This paper presents the diagnosis of ASD based on resting-state eyes-closed EEG signals using machine learning algorithms. The research study population consists of 100 children with ASD (82 male and 18 female) between 5-19 years and 88 healthy developing children in the same range of age with equal proportion of males and female. Power spectrum analysis was used for the analysis of EEG as feature extraction. In addition, feature selection is applied based on principal component analysis for dimensionality reduction. The Logistic Regression, K-Nearest Neighbours, Decision Trees, Random Forest, Extra Trees and Extreme Gradient Boosting classifiers are used for the classification of autistic versus typically developing children. Evaluating the performance of the best classifier among the baseline models results in a classification accuracy of 67.7%, AUC 0.74, 83.3% recall, 61% precision and 54.3% specificity using Extra Trees Additionally, the results revealed that the children with ASD showed convincingly higher power in theta delta and beta bands and low power in alpha than normal group.