Key result
The ST-MEM self-supervised learning framework outperformed baseline methods, achieving an AUROC of 0.933 on the PTB-XL dataset and 0.980 on the CPSC2018 dataset for ECG classification.
Why the study?
Labeled ECG data are limited, and naive self-supervised learning approaches that do not consider spatial-temporal relationships in ECG signals may yield suboptimal representations for disease screening.
Absolute Event Rate: 0.933% vs 0.915%
ST-MEM, a novel spatio-temporal self-supervised learning framework for ECG data, learns robust general representations that outperform existing baseline methods in classifying arrhythmias and myocardial infarction, even in low-resource and reduced-lead settings.
Electrocardiograms (ECG) are widely employed as a diagnostic tool for monitoring electrical signals originating from a heart. Recent machine learning research efforts have focused on the application of screening various diseases using ECG signals. However, adapting to the application of screening disease is challenging in that labeled ECG data are limited. Achieving general representation through self-supervised learning (SSL) is a well-known approach to overcome the scarcity of labeled data; however, a naive application of SSL to ECG data, without considering the spatial-temporal relationships inherent in ECG signals, may yield suboptimal results. In this paper, we introduce ST-MEM (Spatio-Temporal Masked Electrocardiogram Modeling), designed to learn spatio-temporal features by reconstructing masked 12-lead ECG data. ST-MEM outperforms other SSL baseline methods in various experimental settings for arrhythmia classification tasks. Moreover, we demonstrate that ST-MEM is adaptable to various lead combinations. Through quantitative and qualitative analysis, we show a spatio-temporal relationship within ECG data. Our code is available at https://github.com/bakqui/ST-MEM.
No takes yet. Share an insight, caveat, or question.
Na et al. (2024) studied Cardiac Arrhythmia and Myocardial Infarction. ST-MEM (Spatio-Temporal Masked Electrocardiogram Modeling) vs. Other self-supervised learning baselines (e.g., MoCo v3, CMSC, MTAE, MLAE) was evaluated on Arrhythmia and myocardial infarction classification performance (AUROC). The ST-MEM self-supervised learning framework outperformed baseline methods, achieving an AUROC of 0.933 on the PTB-XL dataset and 0.980 on the CPSC2018 dataset for ECG classification.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: