Performance Evaluation of PO, SO and SHO Algorithms for Epileptic Seizer Detection

Diseases originated due to the disorder in the neurons present in the brain part of the body is called epilepsy and the symptom observed in other parts of the body is called a seizure. Not only it is a typical disease in human beings but animals are also suffered from this disease. To live an epilepsy-free life, neurologists and researchers contribute their best effort to devise new techniques and presently metaheuristic optimization algorithms and machine learning algorithms are integrated into this field to design powerful EEG devices for accurately detecting the presence of EEG signals. This research paper presents the comparison of three metaheuristic optimization algorithms used in designing EEG detection. In addition, these three algorithms are integrated with Extreme Learning Adoptive Neuro-Fuzzy-Inference System (ELANFIS) as the classifier and Wavelet Packet Transform as the feature extraction technique. For continuing research work, this experiment uses the Bonn University EEG dataset. After a thorough analysis, it is observed that WPT-PO-ELANFIS is the best-performing technique among these three with average accuracy, sensitivity, specificity, and area under curve values as 98.42, 96.75, 97.97.97.65, and 0.989. This research work also can be tested with other popularly available datasets and integrated with real-time applications.