Classification of Children with Attention Deficit Hyperactivity Disorder Using PCA and K-Nearest Neighbors During Interference Control Task

This study investigates the EEG signals obtained from Children with Attention deficit hyperactivity disorder (ADHD) and typically developing (TD) children while performing a hybrid Simon–spatial Stroop task, which is aimed to achieve a high classification rate. First, a subset EEG channels were selected using principal component analysis (PCA) to preserve as much information as the full set of 128 channels. Second, the feature set consisted of the time-domain amplitude in all the segmentation time windows from 30 subjects with leave-one-out (LOO) cross-validation strategy, which was collected from the optimal channels in prefrontal cortex and inferior parietal area during four different conditions. Then, K-nearest neighbors (K-NN) and support vector machine (SVM) were used to classify ADHD and TD. The results showed that the best classification accuracy of 83.33 % was achieved by K-NN classifier, suggesting that the method could detect ADHD effectively.