Epilepsy Seizures Classification with EEG Signals: A Machine Learning Approach

Abstract

Epilepsy is a neurological disorder characterized by recurrent seizures, which can significantly impact a person's life. Early and accurate diagnosis of epilepsy is crucial for effective management and treatment. The traditional methods for diagnosing epilepsy are deemed ineffective and costly. Epilepsy disease detection at an early stage is crucial. Machine learning techniques have shown promise in automating the classification of epilepsy based on various data sources, such as electroencephalogram (EEG) signals, clinical features, and imaging data. This paper presents a machine learning approach to epilepsy disease classification using EEG signal data. We have applied various machine learning models, including Random Forest, XGBoost, GradientBoost, Naive Bayes, Decision Tree, and Extra Tree, with some pre-processing and feature selection techniques. XGBoost achieved 98.93% training accuracy and 98.23% testing accuracy; Gradient Boost achieved 98.40% training and 98.20% testing accuracy; Extra Tree achieved 98.65% training and 97.85% testing accuracy; Random Forest achieved 97.42% training and 96.52% testing accuracy; Decision Tree achieved 92.6% training and 92.4% testing accuracy; Navies Bayes achieved 93.52% training and 92% testing accuracy. The XGBoost classifier achieved the highest accuracy among all other classifiers applied in the proposed research experiment.