A hybrid model (SVM-LOA) for epileptic seizure detection in long-term EEG records using machine learning techniques

Abstract

The aim of this research is to develop a hybrid Model (SVM-LOA) for epileptic seizure detection with support vector machine (SVM) and lion optimization algorithm (LOA) to locate the optimum parameters of support vector machines (SVMs) for classification of Electroencephalogram (EEG) signals . The proposed approach attempts to find the best integration of all available features that offers typical epilepsy detection and gives a better classification rate. Furthermore, the discrete wavelet transform (DWT) has been implemented to divide EEG signals into five combinations. Nonlinear parameters have been calculated and used to be the features to intern the SVM classifier that affects the classification accuracy. We used lion optimization algorithm based approach to optimize the SVM parameters. The overall experimental results shown the LOA-SVM classifiers have 96.78% for accuracy, but SVM that have obtained 80.05% for accuracy. Therefore; the SVM-LOA is an efficient model for neuroscientists to detect epileptic seizure in EEG.