SPARK: Physiology-informed inference of latent cardiac autonomic dynamics from RR intervals

SPARK reconstructs two latent autonomic drives from RR intervals, reducing solution ambiguity from a median of 23.5 distinct fits to 1.0 when physiological information is included.

• Why it matters: RR intervals alone make parasympathetic-like versus sympathetic-like separation underdetermined, so conventional heart-rate variability analyses cannot uniquely describe autonomic dynamics.
• What they did: The authors built a stochastic phase-domain state-space model with mean-reverting latent drives and integrate-and-fire pacemaker dynamics, using continuous-time Kalman filtering plus time-scale priors, kinetic ordering, and spectral partitioning.
• The result: In recordings fit from many starting points, one process captured essentially all variability in 75% of fits without physiological information versus 4% with it; tilt-table data showed vagal-like decreases and sympathetic-like increases in 10 of 10, and exercise recovery times were 42 s and 341 s.

AI summary of the abstract · may contain errors · not medical advice · read the paper.