Learning from Adaptive Oscillatory Modes: An Interpretable Synchrosqueezing Transformer for Patient-Independent Seizure Detection
4 affiliations
- 1Department of Information Systems, College of Computer and Information Sciences, Jouf University, Sakaka 72388, Saudi Arabia
- 2Department of Computer Science, College of Computer and Information Sciences, Jouf University, Sakaka 72388, Saudi Arabia
- 3Department of Information Technology, College of Computing and Information Technology at Khulais, University of Jeddah, Jeddah 23890, Saudi Arabia
- 4Department of Software Engineering, College of Computer and Information Sciences, Jouf University, Sakaka 72388, Saudi Arabia
Abstract
Transformer-based analysis of physiological signals critically depends on how input tokens are constructed, yet conventional tokens derived from raw temporal windows or spectrogram patches may overlook the intrinsic oscillatory organization of electroencephalographic (EEG) activity. To address this limitation, we propose AMST-Former, an adaptive mode-wise synchrosqueezing Transformer for patient-independent epileptic seizure detection. Each EEG channel is first decomposed into oscillatory modes using Adaptive Local Iterative Filtering (ALIF), after which the Synchrosqueezing Transform (SST) is applied independently to each mode. This decomposition-before-synchrosqueezing strategy reduces cross-component interference in instantaneous-frequency estimation while preserving mode-specific time–frequency dynamics. The resulting time–frequency patches are embedded together with channel, mode-frequency, and positional information. A hierarchical factorized-attention mechanism then models token relevance across time–frequency regions, oscillatory modes, and EEG channels before a Transformer encoder captures their interactions. Evaluation using a repeated patient-independent five-fold cross-validation protocol on the CHB-MIT database (22 subjects) yields an F1-score of 70.76% and an area under the precision–recall curve (AUPRC) of 0.870. At the event level, AMST-Former achieves a sensitivity of 81.4% at an average false detection rate of 1.39 false detections per 24 h. Ablation experiments show that mode-wise SST provides more effective representations than direct SST, per-mode Hilbert spectra, and fixed filter-bank decomposition. Synthetic experiments confirm the cross-component bias predicted by the analysis and shows that the SST of a mode remains accurate when separation is imperfect, whereas its Hilbert instantaneous frequency does not. Furthermore, the factorized-attention mechanism provides mode, time–frequency, and channel-level explanations, whose faithfulness is assessed through a deletion analysis. These results support adaptive mode-wise time–frequency tokenization as a promising strategy for interpretable, patient-independent EEG seizure detection.
Related
- FloREN: Decoding Immune Regulatory Networks through Interpretable Graph Transformer Patient Representations
- Hybrid local–global vision transformer framework for robust plant disease detection in precision agriculture
- Subject-independent classification of ADHD from pediatric EEG: a leakage-aware deep learning framework with interpretable neurophysiological mapping
- An adaptive transformer-based model with entropy-driven knowledge distillation for single-channel EEG REM sleep detection
- CuFormer U-Net: a transformer-enhanced multi-task network for cucumber leaf disease detection and segmentation
BibTeX
@article{alanazi2026learning,
title = {{Learning from Adaptive Oscillatory Modes: An Interpretable Synchrosqueezing Transformer for Patient-Independent Seizure Detection}},
author = {Afrah Alanazi and Kais Khaldi and Sarah Almaghrabi and Hamed Alshammari},
journal = {Mathematics},
year = {2026},
month = oct,
doi = {10.3390/math14213922},
url = {https://www.mdpi.com/2227-7390/14/21/3922},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Learning from Adaptive Oscillatory Modes: An Interpretable Synchrosqueezing Transformer for Patient-Independent Seizure Detection
AU - Afrah Alanazi
AU - Kais Khaldi
AU - Sarah Almaghrabi
AU - Hamed Alshammari
PY - 2026
DA - 2026/10/27
JO - Mathematics
DO - 10.3390/math14213922
UR - https://www.mdpi.com/2227-7390/14/21/3922
AB - Transformer-based analysis of physiological signals critically depends on how input tokens are constructed, yet conventional tokens derived from raw temporal windows or spectrogram patches may overlook the intrinsic oscillatory organization of electroencephalographic (EEG) activity. To address this limitation, we propose AMST-Former, an adaptive mode-wise synchrosqueezing Transformer for patient-independent epileptic seizure detection. Each EEG channel is first decomposed into oscillatory modes using Adaptive Local Iterative Filtering (ALIF), after which the Synchrosqueezing Transform (SST) is applied independently to each mode. This decomposition-before-synchrosqueezing strategy reduces cross-component interference in instantaneous-frequency estimation while preserving mode-specific time–frequency dynamics. The resulting time–frequency patches are embedded together with channel, mode-frequency, and positional information. A hierarchical factorized-attention mechanism then models token relevance across time–frequency regions, oscillatory modes, and EEG channels before a Transformer encoder captures their interactions. Evaluation using a repeated patient-independent five-fold cross-validation protocol on the CHB-MIT database (22 subjects) yields an F1-score of 70.76% and an area under the precision–recall curve (AUPRC) of 0.870. At the event level, AMST-Former achieves a sensitivity of 81.4% at an average false detection rate of 1.39 false detections per 24 h. Ablation experiments show that mode-wise SST provides more effective representations than direct SST, per-mode Hilbert spectra, and fixed filter-bank decomposition. Synthetic experiments confirm the cross-component bias predicted by the analysis and shows that the SST of a mode remains accurate when separation is imperfect, whereas its Hilbert instantaneous frequency does not. Furthermore, the factorized-attention mechanism provides mode, time–frequency, and channel-level explanations, whose faithfulness is assessed through a deletion analysis. These results support adaptive mode-wise time–frequency tokenization as a promising strategy for interpretable, patient-independent EEG seizure detection.
SN - 2227-7390
ER -