Hybrid convolutional neural network and bidirectional LSTM framework for automated epileptic seizure detection supporting sustainable healthcare

Epilepsy is a complex neurological disorder characterized by recurrent seizures, significantly impacting patient quality of life and healthcare systems worldwide. Conventional EEG analysis for seizure detection is manual and labor-intensive, motivating the integration of advanced artificial intelligence techniques. This study proposes a fully data-driven, scalable deep sequential learning framework that combines Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) networks for automated seizure detection from EEG signals. Utilizing the publicly available epileptic seizure recognition dataset, the approach preprocesses raw EEG data using segmentation and Hilbert-Huang Transform for time–frequency representation, converting 1D signals into 2D matrices optimized for CNN spatial feature extraction. The hybrid CNN-BiLSTM model leverages convolutional layers to capture spatial traits and BiLSTM to model bidirectional temporal dependencies, facilitating robust classification of seizure versus non-seizure states. Experimental validation demonstrates superior performance compared to CNN, RNN, LSTM, and BiLSTM baselines, with an accuracy of 99.52%, recall of 98.92%, specificity of 99.67%, and minimal false positive and negative rates. The results indicate that the proposed framework achieves near-perfect automated epileptic seizure detection suitable for potential for clinical application. This research underscores the benefits of integrating spatial and temporal deep learning architectures in EEG analysis and provides a promising pathway toward improved monitoring and early intervention in epilepsy management.