LAEF: A Lead-Agnostic ECG Foundation Model Towards Point-of-Care Diagnostics
Ranking
Overall
76
Content
100
Popularity
19
Observed public metrics from 1 member.
Merged summary
TL;DR - LAEF is a 7M-parameter, lead-agnostic ECG foundation model that processes any subset of ECG leads as a spatiotemporal graph. It improves reduced-lead diagnostics for smartwatches and handheld devices while remaining competitive on full 12-lead ECGs.
- Pre-trained on 9.2M ECGs using masked node modeling and stochastic lead sampling.
- Uses physiologically motivated graph connectivity and attention without zero-padding or architectural changes.
- With one randomly sampled lead, it beat zero-padded alternatives on 17 of 18 datasets; with two leads, on 14 of 18.
- Across 18 datasets, it matched specialized 12-lead baselines more than 12× its size and achieved a mean 3.2-point AUROC gain in reduced-lead settings.
Sources (1)
LAEF: A Lead-Agnostic ECG Foundation Model Towards Point-of-Care Diagnostics
Public signals
Semantic Scholar citations 0 · Semantic Scholar influential citations 0
TL;DR - LAEF is a 7M-parameter, lead-agnostic ECG foundation model that processes any subset of ECG leads as a spatiotemporal graph. It improves reduced-lead diagnostics for smartwatches and handheld devices while remaining competitive on full 12-lead ECGs.
- Pre-trained on 9.2M ECGs using masked node modeling and stochastic lead sampling.
- Uses physiologically motivated graph connectivity and attention without zero-padding or architectural changes.
- With one randomly sampled lead, it beat zero-padded alternatives on 17 of 18 datasets; with two leads, on 14 of 18.
- Across 18 datasets, it matched specialized 12-lead baselines more than 12× its size and achieved a mean 3.2-point AUROC gain in reduced-lead settings.