Multimodal contrastive learning for diagnosing cardiovascular diseases from electrocardiography (ECG) signals and patient metadata
This addresses a practical issue for healthcare facilities with limited ECG equipment, though it is incremental in applying existing contrastive learning methods to this domain.
The paper tackled the problem of diagnosing cardiovascular diseases with fewer ECG leads than the standard 12, using contrastive learning to align embeddings from full and reduced lead signals, which improved diagnostic performance across all lead combinations.
This work discusses the use of contrastive learning and deep learning for diagnosing cardiovascular diseases from electrocardiography (ECG) signals. While the ECG signals usually contain 12 leads (channels), many healthcare facilities and devices lack access to all these 12 leads. This raises the problem of how to use only fewer ECG leads to produce meaningful diagnoses with high performance. We introduce a simple experiment to test whether contrastive learning can be applied to this task. More specifically, we added the similarity between the embedding vectors when the 12 leads signal and the fewer leads ECG signal to the loss function to bring these representations closer together. Despite its simplicity, this has been shown to have improved the performance of diagnosing with all lead combinations, proving the potential of contrastive learning on this task.