Epileptic Seizure Recognition System Using Neural Networks and Support Vector Machine Models

One of the most predominant and hazardous medical conditions is epilepsy. Epileptic seizures can cause direct death of patients. The occurrence of seizures is difficult to detect and is a very important aspect in improving the health of epileptic patients. Thus, it becomes crucial to record the EEG signals and recognise the patterns in the data that result in a seizure. There are many publicly available datasets for research and educational purposes. Previous studies have utilized a variety of methods on these datasets, including neural networks, SVM, XGBoost, ANN, and Random forest classifier. This project focuses on the EEG datasets by the University of California, Irvine (UCI) and the CHB-MIT hospital. We aim on applying the neural networks and SVM algorithms as they are proficient in EEG data analysis. We applied each of the models on both datasets, giving us four models in total for analysis. This research work includes the analysis and comparison of these models and to give an understanding of the working of the methodologies. The research work can be further used by domain experts for the betterment of the existing applications and to improve the overall epilepsy healthcare system.