Abstract: Self-supervised contrastive learning has been proven effective in learning good feature representations by contrasting views through data augmentation, followed by fine-tuning for downstream (classification) tasks, thus finding wide applications. Electrocardiogram (ECG), as a non-invasive, low-risk, and low-cost signal source for cardiovascular diseases, its classification aids in early prevention and precise treatment of conditions like arrhythmia. However, most existing methods for ECG representation learning only perform contrastive learning through temporal perturbation augmentation of examples, overlooking the potential utilization of frequency-domain information, leaving room for further improvement in representation quality. Therefore, a frequency domain mixup augmentation strategy was designed for ECG samples, which generates augmented samples by exchanging frequency domain information between samples to achieve contrastive learning, thus addressing the shortcomings of existing ECG representation learning. In the downstream fine-tuning stage, considering that ECG classification inherently involves a multi-label class imbalance problem, yet there are few targeted methods proposed for it, this paper proposes mitigating this issue by incorporating label frequencies into binary cross-entropy (BCE) loss as logit compensation. Finally, model evaluation was conducted on the CPSC2018 and Chapman datasets. Experimental results demonstrate that integrating the proposed method as an independent module into multiple baseline models improves performance in terms of AUC and mAP metrics. Particularly, significant enhancements were observed in the performance of certain rare disease indicators, thereby validating the effectiveness of this approach.

Key words: Electrocardiogram classification, Arrhythmia, Self-supervised contrastive learning, Multi-label, Class imbalance

摘要： 自监督对比学习通过数据增广视图间的对比已被证明能习得好的特征表征，继而通过微调完成下游（分类）任务，因此得到广泛应用。心电图ECG（Electrocardiogram）作为非侵入、低风险和低成本的心血管疾病常用信号源，其分类有助于早期预防和精确治疗心率失常等。然而现有针对ECG表征学习的大多数方法仅通过对样本进行时域的扰动增广进行对比学习，其忽略了频域潜在的信息利用，留下了进一步提升表征质量的空间，为此，针对ECG样本设计了一个频域mixup的增广策略，通过交换样本间的频域信息生成原始样本的增广实现对比学习，弥补了现有ECG表征学习的不足。在下游微调阶段，考虑到ECG分类本质上属于多标记的类不平衡问题，但仍鲜有针对性方法的提出，为此，提出了结合标签频率对二元交叉熵(BCE)损失作logit补偿缓和该问题。最后在CPSC2018和Chapman数据集上进行模型评估，实验结果表明提出的方法作为独立模块插入至多个基线模型在AUC和mAP指标上均有提高，尤其是个别罕见疾性能指标提升显著，从而验证了该方法的有效性。

关键词: 心电图分类, 心率失常, 自监督对比学习, 多标记, 类不平衡