Comparative study of classifiers for patient specific seizure detection

Automatic seizure detection methods basically decrease the workload of EEG monitoring units. In this study, there is considerable interest in improved offline patient specific approaches because they perform better (High sensitivity & lower false detection rate) than patient-independent ones. In this paper, we present a comparative analysis of different patient specific methods w.r.t different classification models. We consider five patient specific methods, two methods with Gaussian mixture model (GMM), next two methods with Support vector machine (SVM) and one with neural network (NN). We noted that NN based method in compare to the GMM and SVM based method had the best result applied on the same