Beat-SSL reached 93% of an ECG foundation model's performance in multilabel classification and surpassed all other methods in the segmentation task by 4%, despite using only 2.5% of the pretraining data.
Beat-SSL provides an efficient contrastive learning framework for ECG analysis that achieves high performance with significantly less pretraining data.
Obtaining labelled ECG data for developing supervised models is challenging. Contrastive learning (CL) has emerged as a promising pretraining approach that enables effective transfer learning with limited labelled data. However, existing CL frameworks either focus solely on global context or fail to exploit ECG-specific characteristics. Furthermore, these methods rely on hard contrastive targets, which may not adequately capture the continuous nature of feature similarity in ECG signals. In this paper, we propose Beat-SSL, a contrastive learning framework that performs dual-context learning through both rhythm-level and heartbeat-level contrasting with soft targets. We evaluated our pretrained model on two downstream tasks: 1) multilabel classification for global rhythm assessment, and 2) ECG segmentation to assess its capacity to learn representations across both contexts. We conducted an ablation study and compared the best configuration with three other methods, including one ECG foundation model. Despite using 2.5% of the foundation model data on pretraining, Beat-SSL reached 93% of its performance in multilabel classification task and surpassed all other methods in the segmentation task by 4%.
Rizqyawan et al. (2026) studied ECG analysis. Beat-SSL (contrastive learning framework) vs. Three other methods, including one ECG foundation model was evaluated on Multilabel classification for global rhythm assessment and ECG segmentation. Beat-SSL reached 93% of an ECG foundation model's performance in multilabel classification and surpassed all other methods in the segmentation task by 4%, despite using only 2.5% of the pretraining data.