New AI method uses graph neural networks to detect epileptic seizures in EEG signals with improved interpretability

Researchers decompose EEG signals into five frequency bands (delta, theta, alpha, lower beta, higher beta) to analyze brain activity

Framework extracts 11 discriminative features from each frequency band to capture seizure-related patterns

Graph convolutional neural network (GCN) models spatial relationships between EEG electrodes for better detection accuracy

Tested on CHB-MIT scalp EEG dataset, achieving high seizure detection performance with greater neurophysiological relevance than previous deep learning approaches