Abstract
The brain condition known as epilepsy has an impact on patients' quality of life. The need for computer-automated diagnosis systems (CADS) has arisen due to the shortcomings of conventional clinical and machine learning techniques as well as the shortage of neurologists who can identify epilepsy from massive files. With the help of statistical and nonlinear features, this work offers an automated diagnosis system for epileptic seizures based on electroencephalography (EEG) signals that can accurately and rapidly differentiate between seizure stages. The gray wolf optimization (GWO) technique was used to build a feature reduction matrix to reduce computational complexity. A hybrid SVM-Fuzzy machine learning system trained with the goose optimization approach was then used to classify the reduced features. Based on experimental data, the results demonstrate that the suggested approach achieved 98.1% accuracy, 97.8% sensitivity, and 98.4% specificity a notable improvement over current techniques. These discoveries support current initiatives to create automated techniques for diagnosing epilepsy and may result in quicker and more precise diagnoses, which would eventually enhance patient care and results.