Detection of Epileptic Seizures Using ML Algorithms Enhanced by ADASYN and SMOTE Sampling Techniques

Abstract

Machine learning algorithms play a crucial role in healthcare and medical diagnosis applications within the computer-aided research domain. Continuous seizures, which are sudden bursts of electrical activity in the brain, are a hallmark of epilepsy, a neurological condition of the brain. To detect epileptic seizures, this study monitors EEG (Encephalography) signals and turns them into a dataset. Then, using ADASYN (Adaptive Synthetic Sampling) and SMOTE (Synthetic Minority Oversampling Technique) to balance the data, the dataset is subjected to algorithms like Linear Regression, Support Vector Machine, Regressor, Logistic Regression, Decision Tree, KNN (k-nearest neighbor), and Random Forest. This paper investigates the Complete EEG dataset, i.e., Epileptic Seizure Recognition from Kaggle. Based on historical data, these models assist us in identifying epileptic episodes. Every model trained on the dataset produced accurate values. It was recognized that Random Forest with the SMOTE Model gives better accuracy for the given EEG datasets.