Epileptic seizure detection in EEG signals via an enhanced hybrid CNN with an integrated attention mechanism

The model achieved 99.00% accuracy in binary classification of EEG signals.

• A novel deep learning framework was developed that integrates a convolutional neural network, bidirectional gated recurrent unit, and a convolutional block attention module for EEG pattern recognition.
• The framework captures both spatial features and long-term temporal dependencies in EEG data.
• Evaluation on a public EEG dataset demonstrated superior performance compared to existing detection methods.
• The model exhibited high sensitivity ranging from 89.00% to 99.00% and specificity between 89.63% and 99.00% across various classification tasks.
• The findings suggest potential for improved accuracy in diagnosing epileptic seizures.

AI simplified