Measured Structure as Self-Supervision: Phase-Equivariant Latent Dynamics for Cyclic Biosignals

The cyclic structure of physiological processes offers a natural prior for self-supervised representation learning, and the cardiac cycle provides a particularly well-defined setting in which to exploit it. We derive a phase-equivariant self-supervised objective and introduce \Winder{}, a joint-embedding architecture that organises representations into phase-invariant coordinates and phase-rotating harmonic subspaces, via a fixed loss term. Evaluated on PTB-XL under a frozen linear-probe protocol, \Winder{} attains diagnostic accuracy within the range reported by state-of-the-art self-supervised methods at a ~1 M parameter footprint, while exhibiting phase-equivariant latent geometry and improved robustness to single-lead dropout. These results show that measured physiological coordinates provide a testable dynamics prior for reliable health world models without increasing deployed model capacity, enabling fully on-edge, high-throughput $\mathcal{O}(\text{ms})$ condition detection.