Identification of Children with ADHD from EEG Signals Based on Entropy Measures and Support Vector Machine

Attention deficit hyperactivity disorder (ADHD) is one of the major psychiatric and neurodevelopment disorders that affects 11% of children worldwide. Moreover, the prevalence of ADHD has rapidly increased over time worldwide. According to DSM-V, three types of symptoms such as inattention, hyperactivity, and combined type (inattention with hyperactivity). So, it is necessary to use a simple, non-invasive, and automatic detection system for identifying children with ADHD at an early stage. The objective of this study was to propose a machine learning (ML)based ADHD-combine type (ADHD-CT) detection system from electroencephalogram (EEG) signals. EEG signals were recorded from nineteen ADHD-CT children and fourteen healthy children. We extracted five entropy-based features such as approximate-based entropy, Shanon-based entropy, permutation-based entropy, sample-based entropy, and singular value decomposition (SVD)-based entropy from each signal. The subset of the most informative and discriminative features was selected for ADHDCT using sequential forward floating selection (SFFS). Following that, support vector machine (SVM) was implemented with leave-one-out cross-validation for the identification of ADHD-CT children and assessed its performances based on classification accuracy. Our results illustrated that SVM with polynomial kernel provided 96.87% classification accuracy to discriminate children as ADHD-CT and healthy children. Our findings showed that our proposed system can be used to detect children with ADHD-CT.