PaperarXiv:2611.14141
Deviation-Aware Pretraining for Cardiovascular Disease Detection from Electrocardiograms
arXiv preprint (cs.LG)
TL;DR
Self-supervised 12-lead ECG feature extractor pretrained on 918,915 ECGs (MIMIC-IV-ECG + a clinical partner). A codebook of normal beats lets one module localise waveform deviations and another reconstruct cross-lead associations. Reports up to +6.5% AUROC and +12.8% F1 over prior SSL methods with a frozen extractor.
Abstract
Electrocardiogram (ECG) signals are critical for cardiovascular disease (CVD) detection. Most current deep learning models rely on supervised, task-specific training, which limits their ability to generalize across diseases, patient populations, and acquisition settings. This paper presents a self-supervised ECG feature extraction framework that learns from nearly one million unlabeled 12-lead ECG recordings. The framework includes two modules: Deviation Localization Embedding (DLE), which detects waveform-level deviations from normal patterns, and Association Reconstruction Embedding (ARE), which models cross-lead associations through deviation-guided reconstruction to capture abnormal patterns expressed across multiple leads. These modules enable the framework to capture subtle deviation-related patterns and their cross-lead context. The framework achieves up to 6.5% AUROC and 12.8% F1 improvements over prior methods on multiple public datasets, even without additional fine-tuning the pretrained feature extractor. Embedding visualizations show that the learned features support meaningful disease grouping and anomaly identification. The approach reduces the need for labeled data and can support clinicians in diverse healthcare settings. The framework has potential to improve early detection and monitoring of cardiovascular diseases, contributing to more accessible and equitable heart care.
Verbatim from arXiv (opens in a new tab) · arXiv metadata (abstracts are released under CC0)
Results on ChatPPG leaderboards
All leaderboardsNo results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.
Tasks & signals
Signals
Keywords
Datasets used
How it works
Cite this paper
Yan, S., Zheng, H., Lu, F., & Li, W. (2026). Deviation-Aware Pretraining for Cardiovascular Disease Detection from Electrocardiograms. arXiv preprint. https://arxiv.org/abs/2611.14141
Yan, Siyuan, Haibo Zheng, Feng Lu, and Wei Li. 2026. "Deviation-Aware Pretraining for Cardiovascular Disease Detection from Electrocardiograms." arXiv preprint. https://arxiv.org/abs/2611.14141.
Yan, S., Zheng, H., Lu, F. and Li, W. (2026) 'Deviation-Aware Pretraining for Cardiovascular Disease Detection from Electrocardiograms', arXiv preprint. Available at: https://arxiv.org/abs/2611.14141
Yan, Siyuan, et al. "Deviation-Aware Pretraining for Cardiovascular Disease Detection from Electrocardiograms." arXiv preprint, 2026. https://arxiv.org/abs/2611.14141.
S. Yan, H. Zheng, F. Lu, and W. Li, "Deviation-Aware Pretraining for Cardiovascular Disease Detection from Electrocardiograms," arXiv preprint, 2026, [Online]. Available: https://arxiv.org/abs/2611.14141.
@misc{yan2026deviationaware,
title = {{Deviation-Aware Pretraining for Cardiovascular Disease Detection from Electrocardiograms}},
author = {Yan, Siyuan and Zheng, Haibo and Lu, Feng and Li, Wei},
year = {2026},
howpublished = {arXiv preprint},
eprint = {2611.14141},
archivePrefix = {arXiv}
}Added to the ChatPPG research hub
