CLOCS: Contrastive Learning of Cardiac Signals

By Dani Kiyasseh, T. Zhu, David A. Clifton. International Conference on Machine Learning, 2020

317 citations recorded by the source corpus.

The healthcare industry generates troves of unlabelled physiological data. This data can be exploited via contrastive learning, a self-supervised pre-training mechanism that encourages representations of instances to be similar to one another. We propose a family of contrastive learning methods, CLOCS, that encourages representations across time, leads, and patients to be similar to one another. We show that CLOCS consistently outperforms the state-of-the-art approach, SimCLR, on both linear evaluation and fine-tuning downstream tasks. We also show that CLOCS achieves strong generalization performance with only 25% of labelled training data. Furthermore, our training procedure naturally generates patient-specific representations that can be used to quantify patient-similarity.

Subfields: Artificial Intelligence, Cardiology and Cardiovascular Medicine, Pulmonary and Respiratory Medicine

View Paper: https://www.semanticscholar.org/paper/5766878e9d9430f10dbc07fa57876679e0c847dc

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Eric J. Topol
2. Pranav Rajpurkar
3. Mohit Bansal
4. Marinka Zitnik
5. Andrew Y. Ng
6. Percy Liang
7. Theodoros Tsiligkaridis
8. Richard Wang

In Cardiology and Cardiovascular Medicine

1. Eric J. Topol
2. Pranav Rajpurkar
3. Theodoros Tsiligkaridis
4. Marinka Zitnik
5. Andrew Y. Ng
6. Ziyuan Zhao

In Pulmonary and Respiratory Medicine

1. Pranav Rajpurkar
2. Theodoros Tsiligkaridis
3. Marinka Zitnik
4. Andrew Y. Ng
5. Ziyuan Zhao

Citations in the corpus

• MedAug: Contrastive learning leveraging patient metadata improves representations for chest X-ray interpretation (2021, background): “A similar result has been shown by Kiyasseh et al. (2020) for ECG signal interpretation.”
• Self-supervised learning in medicine and healthcare (2022, background): “For example, ECG data (if collected from multiple leads) are both spatially and temporally structured, and in conjunction with data-augmentation techniques have allowed models to generate spatiotemporal associations for the same underlying patient and disease 20,21 .”
• Self-Supervised Contrastive Pre-Training For Time Series via Time-Frequency Consistency (2022, methodology): “Here, the time-based augmentation bank includes jittering, scaling, time-shifts, and neighborhood segments, all well-established in contrastive learning [40, 48, 41].”