Advanced Seizure Detection Framework Using Stacked Convolutional Restricted Boltzmann Machine (SCRBM)

Abstract

Epileptic seizures present major challenges in neurological health, requiring accurate and efficient detection methods for timely diagnosis. This letter presents an advanced seizure detection framework using a stacked convolutional restricted Boltzmann machine (SCRBM) to analyze electroencephalography (EEG) signals. The proposed method integrates convolutional neural networks (CNNs) with restricted Boltzmann machines (RBMs) to effectively capture both spatial patterns and temporal dependencies present in EEG data. Using the Bonn EEG dataset, the model performs remarkably well, achieving 98.7% accuracy, 98.5% sensitivity, and 98.6% precision. A comparison study highlights the benefits of the suggested framework over current techniques, highlighting its applicability, resilience, and effectiveness for real-time epileptic seizure detection. Based on the performance metrics obtained, the application of the stacked CRBM model in clinical settings shows strong potential and effectiveness for real-time epileptic seizure detection.