An Efficient Kernel-SVM-based Epilepsy Seizure Detection Framework Utilizing Power Spectrum Density

Machine learning algorithms can leverage electroencephalogram (EEG) data to extract valuable information. The main objective of this research is to investigate the potential utilization of these technologies in the diagnosis of mental disorders, specifically epilepsy. For feature extraction, the Welch power spectral density (PSD) is used on a dataset from Nigeria that is available through the Zenodo project. The purpose of this is to aid in the diagnosis of epilepsy. Multiple classifiers, including Kernel SVM, Naive Bayes, and Random Forest, are employed for classification. Both methods are used in conjunction with each other. The proposed approach attains a remarkable accuracy of 93.09% by utilizing the Kernel support vector machine (SVM), surpassing other classifier models. The performance results are significant as they have the potential to enhance the diagnosis of neurological disorders, leading to improved patient outcomes.