Cardiac Autonomic Association Atlas of Sleep Parameters in Obstructive Sleep Apnea: A Large-Scale Cohort-Based Observational Study

STUDY OBJECTIVES: To systematically evaluate associations between multidimensional sleep parameters and non-invasive cardiac autonomic nervous system metrics derived from heart rate variability (HRV) in patients with obstructive sleep apnea, and to identify the autonomic pathways most closely linked to sleep disturbances. METHODS: Using overnight polysomnography from the Shanghai Sleep Health Study cohort, a three-stage dimensionality-reduction strategy compressed 147 sleep parameters and 68 HRV metrics into 59 parameters (5 functional domains) and 25 metrics (5 autonomic domains). For each sleep-HRV pair, multivariable linear regression was adjusted for age, sex, BMI, blood pressure, fasting glucose, and lipids. False discovery rate was controlled via the Benjamini-Yekutieli method, with robustness assessed through 2,000 bootstrap resamples and an independent test set. RESULTS: Among 9,359 subjects (training 6,551; test 2,808), 646 of 1,475 associations were significant after correction; 602 (93.2%) showed directional consistency across bootstraps, with 80.3% test-set reproducibility. The Regulatory Complexity and Sympathovagal Balance domains were enriched in Nocturnal Hypoxemia (71.7%, 63.3%) and Fragmentation & Arousal (66.7%, 58.3%), whereas Baroreflex & Ectopic Domain was systematically sparse (6.7% to 20.0%). Core associations arose almost exclusively from Respiratory Event Burden and Nocturnal Hypoxemia, diverging directionally: positive with Lempel-Ziv complexity, skewness, and VLF power, yet negative with Higuchi fractal dimension, normalized corrected conditional entropy. CONCLUSIONS: Nocturnal Hypoxemia and Fragmentation & Arousal are preferentially linked to nonlinear dynamics reflecting Regulatory Complexity and Sympathovagal Balance, whereas Baroreflex & Ectopic metrics are only selectively engaged. This atlas provides a unified framework for generating testable hypotheses and selecting clinically meaningful autonomic biomarkers.