What question did this study set out to answer?

The aim is to improve the detection of epileptic seizures through advanced machine learning and deep learning approaches using EEG signals.

April 5, 2026

Advanced Machine Learning and Deep Learning Approaches for Accurate Epileptic Seizure Detection Using EEG Signals

Key Points

The aim is to improve the detection of epileptic seizures through advanced machine learning and deep learning approaches using EEG signals.
Utilized nine deep learning models including CNN and LSTM for analysis.
Employed seven machine learning models such as XGBoost and Random Forest for classification.
Evaluated models using standard performance metrics, accuracy, and confusion matrices.
CNN achieved 85% accuracy in detecting seizures.
XGBoost outperformed with 88% accuracy in seizure classification.
Demonstrated the effectiveness of ML and DL in identifying patterns from EEG signals.

Abstract

ABSTRACT Epilepsy, a neurological disorder that is caused by improper brain activity that results in seizures, is prevalent in millions of persons across the world and for which diagnosis is extremely crucial for treatment. Early detection of epileptic seizures is paramount as it enables timely intervention, improves quality of life, and prevents potential risks during seizures. Machine learning (ML) and deep learning (DL) algorithms have emerged as powerful tools in revolutionizing medical field, particularly in domains involving images, signals, and other types of visual representations. In our study, we have utilized the capability of these algorithms in exploring their impact on the epileptic seizure detection with the help of EEG Signals. Nine DL models, namely LSTM, GRU, Bi‐LSTM, CNN, FCNN, Hybrid CNN‐LSTM, EEGNet, Shallow ConvNet, and Hybrid CNN‐GRU and seven machine learning models that is, Random Forest, XGBoost, KNN, Logistic Regression, Naïve Bayes, Decision Tree, and Stacking Ensemble are employed for classification of epileptic seizures on EEG signal dataset. All the models have been evaluated using the standard performance metrics for determining the effectiveness of these models on the epileptic seizure classification as epileptic and nonepileptic. Visual representations like accuracy‐loss graphs and confusion matrix were also generated for better visual understanding of the performance of the model. Among all DL Models, CNN emerged as the best performing model with 85% accuracy, whereas, among ML models, XGBoost performs best with accuracy of 88%. The study underscores the potential and effectiveness of ML and DL models in detecting complex patterns and generating predictive insights for classification and detection of epileptic seizure from EEG signals.

Perguntar à IA

Bookmark