Yogatheesan Varatharajah, Ahmed Sameh, Ramzi Al-Sharawi
We lifted 8 functions out of this paper's own repositories and ran 7 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| muha-0/ecg-ssl-representation-learning | canonical | 7 of 8 |
| Function | Status | Where it lives |
|---|---|---|
| average_precision | Ran | muha-0/ecg-ssl-representation-learning/eval/metrics.py code served (permissive licence) · get_code("61b235ee7c0e69cc") |
| compute_metrics | Ran | muha-0/ecg-ssl-representation-learning/eval/metrics.py code served (permissive licence) · get_code("140160542a17a7b4") |
| embed_probe_set | Ran | muha-0/ecg-ssl-representation-learning/eval/convergence.py code served (permissive licence) · get_code("1ce9ae6a8c5847ee") |
| get_tsne_embeddings | Ran | muha-0/ecg-ssl-representation-learning/eval/t_sne.py code served (permissive licence) · get_code("d5a48d160133173d") |
| info_nce | Ran | muha-0/ecg-ssl-representation-learning/models/losses.py code served (permissive licence) · get_code("73714c075bb607a4") |
| load_offline_codebook | Ran | muha-0/ecg-ssl-representation-learning/codebooks/build_codebook.py code served (permissive licence) · get_code("f137c5dc0b1e879e") |
| recall_at_k | Ran | muha-0/ecg-ssl-representation-learning/eval/convergence.py code served (permissive licence) · get_code("df1982d38efa426b") |
| auc_roc | Not yet run | muha-0/ecg-ssl-representation-learning/eval/metrics.py code served (permissive licence) · get_code("58a4de22ca48c55a") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Self-supervised electrocardiogram (ECG) models are often trained on a few seconds of ECG signal and, increasingly, on discretized token sequences. It remains unclear whether these choices sacrifice information needed for rhythm inference and longitudinal consistency in real-world ambulatory recordings. We present a controlled study on the Icentia11k single-lead dataset that varies (i) the input horizon (16 seconds, 1 minute, 5 minutes, and 10 minutes) and (ii) the front-end representation (continuous convolutional patch embeddings vs. fixed vector-quantized tokens), while holding the Transformer backbone and training protocol constant. Representations are assessed by downstream abnormal rhythm detection and by patient-level retrieval that probes cross-session stability. Our results show that increasing temporal context beyond 16-second snapshots yields stronger transfer and higher retrieval accuracy, with the strongest performance achieved by the 5-and 10-minute models, indicating improved capture of slow-varying rhythm dynamics and individual-specific structure. Across all evaluated horizons, continuous patch embeddings outperform discretized tokens, suggesting that quantization can discard clinically relevant waveform detail. These findings motivate ECG foundation models that emphasize extended context and continuous encoders for clinical prediction and similarity-based applications. Our code and pretrained models are publicly available at https: //github.com/muha-0/ecg-ssl-representation-learning.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2608.12695")
get_code_for_paper("2608.12695")
have("2608.12695")
Connect an agent — have() is free.