Research map / Label-efficient and robust learning / Self-supervised and contrastive
Tracing the Heart’s Pathways: ECG Representation Learning from a Cardiac Conduction Perspective
Tan Pan, Yixuan Sun, Chen Jiang, Qiong Gao, Rui Sun, Xingmeng Zhang, Zhenqi Yang, Limei Han, Yixiu Liang, Yuan Cheng, Kaiyu Guo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add conduction-view sparse attention pretraining (CLEAR) instead of MAE (ablation) | linear probing · PTB-XL Rhythm (1% train data) | macro AUC: 73.73 -> 79.24 | Improved |
| use CLEAR pretraining instead of MAE baseline (ablation) | linear probing · PTB-XL Subclass (1% train data) | macro AUC: 67.49 -> 73.86 | Improved |
| add pre-training (MAE baseline vs scratch) (ablation) | MAE baseline · PTB-XL Rhythm (1% train data) | macro AUC: 51.51 -> 73.73 | Improved |
| add HUG head (ablation) | MAE baseline · PTB-XL Rhythm (1% train data) | macro AUC: 73.73 -> 73.76 | No clear effect |
| add HUG head (ablation) | MAE baseline · PTB-XL Subclass (1% train data) | macro AUC: 67.49 -> 75.41 | Improved |
| add HUG head (ablation) | CLEAR · PTB-XL Subclass (1% train data) | macro AUC: 73.86 -> 76.66 | Improved |
| add HUG head (ablation) | CLEAR · CSN (1% train data) | macro AUC: 62.88 -> 72.09 | Improved |
| use hierarchical grouping instead of single-level grouping (ablation) | HUG · CSN (1% train data) | macro AUC: 68.70 -> 72.09 | Improved |
| use hierarchical grouping instead of single-level grouping (ablation) | HUG · CSN (100% train data) | macro AUC: 89.35 -> 89.93 | Improved |
| use CLEAR (pretraining only) instead of HeartLang | linear probing · PTB-XL Superclass (10% train data) | macro AUC: 85.59 -> 85.43 | Hurt |
| use CLEAR-HUG instead of HeartLang | linear probing · PTB-XL Rhythm (1% train data) | macro AUC: 62.08 -> 79.48 | Improved |
Closest papers by meaning
- From Token to Rhythm: A Multi-Scale Approach for ECG-Language Pretraining (ICML 2025 · similarity 0.89)
- Reading Your Heart: Learning ECG Words and Sentences via Pre-training ECG Language Model (ICLR 2025 · similarity 0.89)
- Leveraging an ECG Beat Diffusion Model for Morphological Reconstruction from Indirect Signals (NeurIPS 2024 · similarity 0.88)
- Parallel-Learning of Invariant and Tempo-variant Attributes of Single-Lead Cardiac Signals: PLITA (AAAI 2025 · similarity 0.88)
- Guiding Masked Representation Learning to Capture Spatio-Temporal Relationship of Electrocardiogram (ICLR 2024 · similarity 0.88)
- GEM: Empowering MLLM for Grounded ECG Understanding with Time Series and Images (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
