Decentralized AI Framework for Privacy-Preserving Epileptic Seizure Detection Using EEG Signals

Epilepsy is a neurological disorder characterized by recurrent seizures caused by abnormal brain activity, requiring continuous monitoring for effective diagnosis. Electroencephalography (EEG) is widely used for seizure detection; however, EEG signals are complex, non-stationary, and prone to noise, making manual analysis difficult. This paper proposes an automated end-to-end deep learning framework for epileptic seizure detection using multi-channel EEG signals. The proposed model, ConvSeizureNet, is a one-dimensional Convolutional Neural Network (1D-CNN) designed to extract discriminative temporal features directly from raw EEG data without handcrafted feature engineering. A robust preprocessing and segmentation pipeline is employed to enhance signal quality. The framework is evaluated on the CHB-MIT dataset using standard metrics and Leave-One-Patient-Out cross-validation. Experimental results achieve 99.13% accuracy, 98.55% precision, and 99.88% specificity. The findings demonstrate strong detection capability, while highlighting challenges in cross-subject generalization for real-world clinical deployment.