Multimodal Epileptic Seizure Detection System (SeizeIT2)
Published:
Duration: 2024 – 2025
Participated in the EU multi-center clinical project SeizeIT2, with 11,640 hours of monitoring data from 125 epilepsy patients. The dataset contains four modalities: EEG, ECG, EMG, and accelerometer data. Developed attention-based multimodal deep learning algorithms with cross-modal feature fusion strategies, improving seizure detection accuracy and robustness. Explored model lightweight techniques for edge device deployment.