Reproduce “Fusion of ECG Foundation Model Embeddings to Improve Early Detection of Acute Coronary Syndromes” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zeyuan Meng, Lovely Yeswanth Panchumarthi, Saurabh Kataria, Alex Fedorov, Jessica Zègre-Hemsey, Xiao Hu, Ran Xiao
This research explores combining embeddings from two distinct ECG foundation models, ST-MEM and ECG-FM, to improve Acute Coronary Syndrome (ACS) detection in pre-hospital settings. The fusion approach achieved an AUROC of 0.843 and AUCPR of 0.674 for ACS detection, outperforming individual foundation models and a ResNet-50 baseline.
Key insights
- Different self-supervised learning strategies, such as reconstruction (ST-MEM) and contrastive learning (ECG-FM), capture distinct and complementary information from ECG signals.
- Fusing embeddings from multiple ECG foundation models enhances diagnostic performance for Acute Coronary Syndrome (ACS) detection, leveraging the complementary nature of their representations.
- Self-supervised pre-training in ECG foundation models generates effective representations that outperform traditional supervised deep learning methods for downstream tasks like ACS detection.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.