Beat-synchronous ECG tokenizers (UMN, IEEE MLSP 2026)
- Form factor
- Chest clinical 12lead and single lead
Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.
Transformer encoders that tokenize ECG at cardiac-cycle (beat) boundaries instead of fixed time patches, from the Beat-Synchronous Tokenization for ECG Transformers paper (arXiv 2608.30367, IEEE MLSP 2026). The GitHub v1.0 release ships pretrained SSL encoders and fine-tuned classifiers for the three beat-synchronous tokenizers (tok1/tok2/tok3) and the fixed-patch baselines (fixed50/100/250/500) on PTB-XL, plus Icentia11k encoders (icentiabeatsync, icentiabeatsynchr, icentiafixed, icentiacontrastive). The README does not map tok1–tok3 to the paper's three designs; check before use.
- Form factor
- chest_clinical_12lead_and_single_lead
- Signals
- ECG
- License
- MIT (repo)
Key results
All results- Beat-synchronous tokenization (Tok1)Table 1 (source, opens in a new tab)
0.8945
AUROC ↑
ECG diagnosis · PTB-XL (5 superclasses)
5 diagnostic superclasses, macro AUROC · transformer pretrained on MIMIC-IV-ECG, evaluated on PTB-XL
Results as reported by the authors; not independently reproduced. A SOTA star marks the best value among results reported under the same evaluation protocol; ties go to the earlier publication.
- Type
- Task model
- Modalities
- ECG
- Form factor
- Chest clinical 12lead and single lead
- Parameters
- Not reported
- Architecture
- Not reported
- Venue
- Not reported
- Year
- Not reported
- arXiv
- Not reported
- DOI
- Not reported
- Code
- Open
- Weights
- Open
- Caveats
- Not reported
- Datasets
- Not reported
- Added
- Oct 17, 2026
- Updated
- Oct 17, 2026
How it works
- Inputs
- ECG
- Approach
- ECG · tokenization · ssl · masked reconstruction · contrastive · PTB XL
Datasets
Papers· 1
Authors· 4
Cite
APA
Beat-synchronous ECG tokenizers (UMN, IEEE MLSP 2026). (n.d.).
BibTeX
@misc{beatsyncecgtokenbeatsynchronous,
title = {{Beat-synchronous ECG tokenizers (UMN, IEEE MLSP 2026)}}
}Built from the hub record. Check it against the publisher before submitting.
History
Added to the ChatPPG research hub
Connections
Open in the graph explorer- Datasets 1
- PTB-XL
- Signals 1
- ECG
- Authors 4
- Ahmed Sameh, Yogatheesan Varatharajah, Max Enderlein, Nolan Wilson
Papers
Beat-Synchronous Tokenization for ECG TransformersPaper on this modelIEEE MLSP 2026 · 2026
The Impact of Temporal Context Length and Encoding Strategies on Self-Supervised ECG Representation LearningShared author: Ahmed SameharXiv preprint · 2026
Bio-PM: Bio-Inspired Self-Supervised Learning for Wrist-worn Accelerometer DataSame task: Signal denoisingarXiv preprint · 2026
