EEG-Titans: Long-Horizon Seizure Forecasting via Dual-Branch Attention and Neural Memory
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908776133558272 |
|---|---|
| author | Pham, Tien-Dat Tran, Xuan-The |
| author_facet | Pham, Tien-Dat Tran, Xuan-The |
| contents | Accurate epileptic seizure prediction from electroencephalography (EEG) remains challenging because pre-ictal dynamics may span long time horizons while clinically relevant signatures can be subtle and transient. Many deep learning models face a persistent trade-off between capturing local spatiotemporal patterns and maintaining informative long-range context when operating on ultralong sequences. We propose EEG-Titans, a dualbranch architecture that incorporates a modern neural memory mechanism for long-context modeling. The model combines sliding-window attention to capture short-term anomalies with a recurrent memory pathway that summarizes slower, progressive trends over time. On the CHB-MIT scalp EEG dataset, evaluated under a chronological holdout protocol, EEG-Titans achieves 99.46% average segment-level sensitivity across 18 subjects. We further analyze safety-first operating points on artifact-prone recordings and show that a hierarchical context strategy extending the receptive field for high-noise subjects can markedly reduce false alarms (down to 0.00 FPR/h in an extreme outlier) without sacrificing sensitivity. These results indicate that memory-augmented long-context modeling can provide robust seizure forecasting under clinically constrained evaluation |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_13748 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | EEG-Titans: Long-Horizon Seizure Forecasting via Dual-Branch Attention and Neural Memory Pham, Tien-Dat Tran, Xuan-The Machine Learning Human-Computer Interaction Accurate epileptic seizure prediction from electroencephalography (EEG) remains challenging because pre-ictal dynamics may span long time horizons while clinically relevant signatures can be subtle and transient. Many deep learning models face a persistent trade-off between capturing local spatiotemporal patterns and maintaining informative long-range context when operating on ultralong sequences. We propose EEG-Titans, a dualbranch architecture that incorporates a modern neural memory mechanism for long-context modeling. The model combines sliding-window attention to capture short-term anomalies with a recurrent memory pathway that summarizes slower, progressive trends over time. On the CHB-MIT scalp EEG dataset, evaluated under a chronological holdout protocol, EEG-Titans achieves 99.46% average segment-level sensitivity across 18 subjects. We further analyze safety-first operating points on artifact-prone recordings and show that a hierarchical context strategy extending the receptive field for high-noise subjects can markedly reduce false alarms (down to 0.00 FPR/h in an extreme outlier) without sacrificing sensitivity. These results indicate that memory-augmented long-context modeling can provide robust seizure forecasting under clinically constrained evaluation |
| title | EEG-Titans: Long-Horizon Seizure Forecasting via Dual-Branch Attention and Neural Memory |
| topic | Machine Learning Human-Computer Interaction |
| url | https://arxiv.org/abs/2601.13748 |