The Respiration–Limb Coactivation Index (RLCI): A Continuous Signal-Based Biomarker for Stratifying Cardiovascular Disease Risk in a Cross-Sectional Sleep Cohort

STUDY OBJECTIVES: Conventional respiratory-related limb movement (RRLM) metrics rely on event definitions and overlap rules that miss continuous respiratory-motor coupling intensity. We developed the respiration-limb coactivation index (RLCI), an algorithmic continuous biomarker, and evaluated associations with cardiovascular disease (CVD). METHODS: In Human Sleep Project polysomnography, RLCI was computed in S0001 (N = 20 737) as the interaction between respiratory-flow recovery and phasic leg-EMG envelope. The locked S0001 pipeline, without parameter tuning, was applied to two independent HSP centers (I0004, I0006). Prevalent coronary heart disease (CHD), cerebrovascular disease (CER), and peripheral vascular disease (PVD) associations were assessed by multivariable logistic regression with stage-restricted comparisons against LMI, PLMI, and RRLMI and signal-processing sensitivity analyses. RESULTS: After covariate adjustment, whole-night RRLMI was not significant for CHD or CER, whereas RLCI metrics, especially in N1 and first-half sleep, remained associated with CER and PVD. RLC N1 showed the strongest CER association (per-1-SD OR 1.11, 95% CI 1.06-1.15), and RLC FirstHalf was associated with PVD (OR 1.07, 95% CI 1.03-1.11). Stage-restricted RRLMI_N1 was significant for all four outcomes, supporting stage-matched benchmarking. Mutual adjustment suggested complementary RLCI information for CER and PVD, while conventional N1 respiratory-related movements better captured CHD. Findings were stable in sensitivity analyses and external validation. CONCLUSIONS: RLCI associations with CER and PVD suggest respiratory-linked motor coactivation captures autonomic surges distinct from cumulative hypoxia or sleep fragmentation and may complement AHI-based cardiovascular risk stratification.