Developing a Robust Model to Detect Attention Deficit Hyperactivity Disorder Using LS-SVM and EEG brain signals

Attention Deficit Hyperactivity Disorder (ADHD) is the term used to describe the most common psychiatric disease in children today. One of the most significant challenges to this disorder’s management and treatment is early diagnosis among children with ADHD. Electroencephalography (EEG) plays a crucial role in evaluating the neurophysiology of ADHD because of its capability to reveal complex brain activity. This study included 45 ADHD patients and 45 healthy and suggested a new model for ADHD detection using frequency domain features combined with a least square support vector machine (LS-SVM). The statistical features were tested to find the optimal feature sets based on K-means that were extracted using discrete wavelet transform (DWT). Putting the selected features into two classifiers, LS-SVM and Naive-Bayes. The performance of the proposed model was evaluated by 10-fold cross-validation, specificity metrics, and other precision sensitivity. 19 channels’ results and the mean accuracies were recorded and calculated. The proposed model obtained an average accuracy of 96.49% with LS-SVM, while it recorded 6S.39% with the Naive-Bayes classifier that is used to classify ADHD.