MS-ECG-FM
multi-source contrastive ECG foundation model (Apple)
ECG
Key facts
- Pretraining Data
- MIMIC-IV-ECG train split (720,102 of 800,035 10 s 12-lead ECGs, 500 Hz) contrastively aligned to five report types: ECG machine reports, ECG cardiologist reports, echo reports, chest X-ray reports, discharge summaries
- Venue
- arXiv preprint 2026-10-06 (cs.LG); 52 pp
Summary
A transformer ECG encoder with a deep shared per-lead convolutional tokenizer (0.5 s patches, 12 × 20 = 240 tokens for a 10 s 12-lead ECG), learnable lead embeddings and 1D sinusoidal time encoding. It is built to transfer to arbitrary lead subsets (Lead I-only is called out as the wearable case). Instead of aligning ECGs only to machine interpretation reports (MERL, D-BETA, MELP), it contrastively aligns each ECG to five clinical text sources: ECG machine reports, cardiologist ECG reads, echo reports, chest X-ray reports and discharge summaries.
Reported benchmarks
AFib / arrhythmia
ptb-xl-rhythm · AUROC ↑
#1 of 1
0.987
PTB-XL Rhythm AFIB label, 12-lead, linear probe, 100% labels (Table 3 / D1); MIMIC-IV-ECG pretrain