Optimized Seizure Detection using Classical Machine Learning Models on EEG Signals

Electroencephalogram (EEG) based epileptic seizure detection is a critical area of research with significant implications for patient care and clinical decision-making. This study proposes an in-depth framework that leverages classical machine learning (ML) models for the robust and real-time detection of seizure events, the purpose of which is to address the need for efficient and lightweight alternatives to deep learning-based methods. The methodology involves the use of a dataset that has been extensively preprocessed with steps such as noise reduction, feature scaling, and class balancing. Then, the comparison of multiple ML classifiers, including popular classifiers such as Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), Gradient Boosting (GB), and Support Vector Machine (SVM). Through rigorous experimentation and evaluation using multiple performance metrics (accuracy, precision, recall, F1-score, and ROC curve), the results demonstrate that ensemble methods (such as RF and GB) consistently outperform simpler models, achieving an accuracy level exceeding 96% while still maintaining and proper balance between precision and recall. In addition to quantitative results, the study includes extensive visualizations such as confusion matrices and ROC curves that detail model performance, discusses the challenges of EEG data processing, and provides insights into the practical deployment of these models in real-time clinical environments. Overall, the findings confirm that classical ML models, when properly tuned and optimized, provide a computationally efficient and highly accurate alternative for seizure detection, with potential applications in portable monitoring systems and resource-constrained settings.