• Abstract Cardiovascular diseases remain a major contributor to the global burden of healthcare, highlighting the importance of accurate and scalable methods for cardiac monitoring. • Cardiac biosignals, most notably electrocardiograms (ECG) and photoplethysmograms, are essential for diagnosing, preventing and managing cardiovascular conditions across clinical and home settings. • However, their acquisition varies substantially across scenarios and devices, whereas existing analytical models often rely on homogeneous datasets and static bespoke models, limiting their robustness and generalizability in diverse real-world contexts. • Here we present a cardiac sensing foundation model (CSFM) that leverages transformer architectures and a generative masked pretraining strategy to learn unified representations from heterogeneous health records. • CSFM is pretrained on a multimodal integration of data from various large-scale datasets, comprising cardiac signals from approximately 1.7 million individuals and their corresponding clinical or machine-generated text reports. • The embeddings derived from CSFM act as effective, transferable features across diverse cardiac sensing scenarios, supporting a seamless adaptation to the varied input configurations and sensor modalities.

Article Summaries:

  • A new cardiac‑sensing foundation model (CSFM) has been trained on multimodal data from about 1.7 million people, combining electrocardiograms (ECGs) and photoplethysmograms (PPGs). Using a transformer architecture and generative masked‑pretraining, CSFM learns unified representations that can be transferred across diverse sensing scenarios. In tests on diagnosis, demographic inference, vital‑sign estimation, outcome prediction and ECG question answering, it consistently outperforms single‑modal, task‑specific models. Importantly, CSFM maintains strong performance on both 12‑lead and single‑lead ECGs, as well as on ECG‑only, PPG‑only, or combined inputs, underscoring its potential as a versatile, scalable foundation for cardiac monitoring.

Sources: