Epileptic EEG Signals Classification Based on Multi-Criteria Decision Aid Classifier Ensemble Approach

Epilepsy is a persistent neurological condition that impacts a significant portion of the worldwide population. Accurate identification is necessary in order to efficiently diagnose, treat, and manage epilepsy. Electroencephalography (EEG), a clinically non-invasive technique, is predominantly utilized for the purpose of monitoring these abnormalities in order to achieve an accurate diagnosis of the ailment. Nevertheless, the process of visually evaluating EEG data is laborious and requires a significant amount of time. Therefore, the implementation of an automated prediction model that employs a machine learning-based technique can successfully overcome these restrictions. However, these predictions may not always be accurate, and they often result in incorrect classifications. To address this challenge, we have applied an ensemble classification approach that offers a comprehensive evaluation, which will enhance the model's accuracy. First, the discrete wavelet packet method has been utilized to decompose the EEG signal into sub-bands, from which we retrieve time-frequency based statistical and entropy characteristics of the data. In the next stage, for developing an ensemble classifier for EEG signals in epileptic patients, we use a multi-criteria decision making (MCDM) methodology by as-sessing various performance measures of the EEG classifiers like sensitivity, specificity, and F1-score, providing a comprehensive evaluation. Till now, no work has been done where such a MCDM based classifier ensemble approach is applied for EEG signal classification. The proposed approach has been evaluated using the publicly available Bonn and TUEP datasets. It has achieved an accuracy of 100%, 83% in the Bonn binary and multiclass datasets, and 98 % in the TUEP dataset. These results have been compared with the state-of-the-art methods. It has been observed that our proposed approach surpasses existing models in the literature.