Effectiveness of TLBO, SCA and WaOA for Epileptic Seizer Detection

Epilepsy is a brain-related disease caused due to imbalance of electricity flow in the brain. During the past two decades, researchers and neurologists have been trying their best to devise powerful EEG (electroencephalogram) devices. In the present scenario, metaheuristic optimization algorithms are integrated with machine learning algorithms to use in the seizure detection symptoms. This research studies three metaheuristic algorithms named teaching learning-based optimization algorithm (LBO), sine cosine algorithm (CSA), and Walrus Optimization algorithm (WaOA) with Wavelet Packet transform as the feature extraction technique, Analysis of Variance (ANOVA) as the feature selection technique and Least square SVM (LSSVM) as the classifier and concludes Walrus Optimization algorithm as the best among the three algorithms having average accuracy, sensitivity, specificity and area under curve as 98.52, 97.54, 93.88, and 0.98. To continue research, it uses the Bonn University EEG dataset. In the future, this result can be tested with other popularly used datasets and also integrated with the Internet of Medical Things (IoMT).