Machine and deep learning methods for epileptic seizure recognition using EEG data: A systematic review

Epilepsy is a neurological disorder affecting millions worldwide, characterized by recurrent and unpredictable seizures. Electroencephalography (EEG) is a widely used tool for seizure diagnosis, but the complexity and variability of EEG signals make manual analysis challenging. Machine Learning (ML) and Deep Learning (DL) techniques have emerged as powerful methods for automated Epileptic Seizure (ES) detection, classification, and prediction. However, questions remain regarding their effectiveness, interpretability, and clinical applicability. This systematic review critically examines ML and DL approaches applied to EEG-based seizure recognition, highlighting key challenges such as feature extraction, dataset selection, and model generalization. We analyze peer-reviewed studies from 2013 to 2023, sourced from the PubMed database, to compare various methodologies and evaluate their performance. Unlike prior reviews that focus on a single aspect of seizure recognition, this work provides a comprehensive overview of detection, classification, and prediction tasks. We also discuss the strengths and limitations of different ML and DL models, emphasizing the trade-offs between computational complexity, accuracy, and real-world implementation. Furthermore, this study outlines emerging trends, including the integration of explainable AI, transfer learning, and privacy-preserving techniques such as federated learning. By synthesizing the latest advancements, this review serves as a guide for researchers and clinicians seeking to enhance the reliability and efficiency of seizure recognition systems. Our findings aim to bridge the gap between AI-driven methodologies and clinical applications, paving the way for more robust and interpretable ES detection frameworks.