State-Transition Dynamics in the HR--Deceleration Capacity Feature Space: Passive Assessment of Autonomic Dysfunction from 24-Hour Heart Rate

Autonomic dysfunction is a critical predictor of mortality in aging, Type 2 Diabetes (T2DM), and Congestive Heart Failure (CHF). Active clinical tests such as the Valsalva maneuver and tilt table impose procedural burdens, while standard HRV metrics cannot distinguish transient physiological states from underlying regulatory function. We propose HR-DC State-Transition Dynamics, a passive framework that maps 24-hour RR intervals onto a Heart Rate vs. Deceleration Capacity (HR-DC) feature space and partitions it into four physiological states using a Gaussian Mixture Model. An Onset-Track-Stabilize protocol then extracts directed state-transition trajectories, yielding three biomarkers (transition efficiency , directional complexity H, and tortuosity {tau}) for both Relaxation and Excitation streams. In a longitudinal pilot (N=10, recordings on Monday/Wednesday/Saturday), the three metrics showed high test-retest reliability (ICC >0.85) across days with markedly different activity levels. In a six-group cross-sectional cohort (N=209), an SVM distinguished healthy elderly from pathological subjects with AUC = 0.85, and the metrics correlated with age (R2=0.33). In a clinical subset (N=26) with concurrent active reflex testing, Relaxation metrics predicted Valsalva Ratio (R2=0.58) and Excitation metrics predicted Tilt SBP drop (R2=0.62), indicating that the framework can serve as a passive surrogate for standard autonomic reflex tests.