CuPID: Leveraging Masked Single-Lead ECG Modelling for Enhancing the Representations
Keywords: Self-Supervised Learning, Time Series, ECG
TL;DR: We present CuPID, a novel SSL method for Single-Lead ECG processing which deals with the idiosyncrasies of this kind of data and outperforms state-of-the-art methods in a variety of downstream tasks and databases.
Abstract: Wearable sensing devices, such as electrocardiogram (ECG) heart-rate monitors, will play a crucial role in the future of digital health. This constant monitoring leads to massive unlabeled datasets, making the development of unsupervised learning frameworks essential to associate these single-lead ECG signals with their anticipated clinical outcomes. While Masked Data Modelling (MDM) methods have enjoyed wide use, the idiosyncrasies of single-lead ECG data make its direct application impractical. In this paper, we present Cueing the Predictor Increments the Detailing (CuPID), a novel Self-Supervised Learning (SSL) method that adapts MDM methods for use on single-lead ECG signal data. CuPID accomplishes this via cueing spectrogram-derived context to the predictors, thus incentivizing the encoder to produce more detailed representations. This leads the class token to accommodate fine-grained information. We demonstrate that CuPID outperforms state-of-the-art methods in a variety of downstream tasks and databases, increasing the accuracy for each task from 3.6 % to 9.7%.
Supplementary Material: zip
Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 8117
Loading