An Optimized Hybrid CNN-LSTM Model for Epileptic Seizure Detection and Prediction

Timely detection and prediction of epileptic seizures are critical for enabling rapid clinical intervention. Conventional Εlectroencephalogram (EEG) analysis, however, is labor-intensive and prone to inaccuracies, highlighting the need for automated solutions. This study proposes an optimized hybrid Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM) model that enhances seizure detection by integrating spatial feature extraction (CNN) with temporal pattern recognition (LSTM). The model was trained and validated using the publicly available CHB-MIT EEG dataset, with performance further improved through hyperparameter optimization and feature selection. Experimental results show that the hybrid model achieves an accuracy of 98.5%, outperforming standalone CNN (95.8%) and LSTM (94.2%) models. Moreover, the proposed hybrid model achieves a False Positive Rate (FPR) of only 1.06%, surpassing the individual CNN (5.32%) and LSTM (4.26%) models. These findings demonstrate the potential of the proposed hybrid model in real-time monitoring epileptic episodes application.