Beyond Patient Invariance: Learning Cardiac Dynamics via Action-Conditioned JEPAs

TL;DR

This paper introduces a novel action-conditioned world model for cardiac time-series analysis that captures disease progression dynamics, outperforming static models especially in low-resource settings.

Contribution

It adapts the LeJEPA framework to physiological data, modeling pathology as dynamic state transitions rather than static labels, improving robustness and sample efficiency.

Findings

Outperforms supervised baselines on triage tasks.

Demonstrates over 0.05 AUROC improvement in low-resource regimes.

Effectively disentangles anatomical features from pathological dynamics.

Abstract

Self-supervised learning in healthcare has largely relied on invariance-based objectives, which maximize similarity between different views of the same patient. While effective for static anatomy, this paradigm is fundamentally misaligned with clinical diagnosis, as it mathematically compels the model to suppress the transient pathological changes it is intended to detect. We propose a shift towards Action-Conditioned World Models that learn to simulate the dynamics of disease progression, or Event-Conditioned. Adapting the LeJEPA framework to physiological time-series, we define pathology not as a static label, but as a transition vector acting on a patient's latent state. By predicting the future electrophysiological state of the heart given a disease onset, our model explicitly disentangles stable anatomical features from dynamic pathological forces. Evaluated on the MIMIC-IV-ECG dataset, our approach outperforms fully supervised baselines on the critical triage task. Crucially, we demonstrate superior sample efficiency: in low-resource regimes, our world model outperforms supervised learning by over 0.05 AUROC. These results suggest that modeling biological dynamics provides a dense supervision signal that is far more robust than static classification. Source code is available at https://github.com/cljosegfer/lesaude-dynamics