Assessment of Heart Rate Variability for Predicting Obstructive Sleep Apnea in Patients With Hypertension

Hypertension, Obstructive Sleep Apnea

Heart Rate Variability, Polysomnography, CPAP, Sleep Apnea Screening, Machine Learning

This study aims to investigate the relationship between hypertension and obstructive sleep apnea (OSA) using heart rate variability (HRV) as a non-invasive biomarker of autonomic nervous system function. Hypertensive patients at high risk for OSA will undergo 24-hour Holter electrocardiogram monitoring to assess HRV parameters, along with overnight polysomnography (PSG) to determine OSA severity. The study will analyze the association between HRV indices and the apnea-hypopnea index (AHI), and develop predictive models using machine learning techniques. In addition, patients diagnosed with OSA will be followed after treatment, and changes in HRV and blood pressure will be evaluated to assess treatment effects and autonomic function recovery. The results of this study may provide a cost-effective and clinically applicable approach for early detection, risk stratification, and management of OSA in patients with hypertension.

This study is designed as a prospective observational cohort study to investigate the association between heart rate variability (HRV), hypertension, and obstructive sleep apnea (OSA). Participants will be recruited from outpatient clinics and include adult patients aged 18 to 65 years with diagnosed hypertension or resistant hypertension and a STOP-Bang score ≥3. After providing informed consent, eligible participants will undergo 24-hour ambulatory electrocardiogram monitoring using a Holter device to obtain HRV parameters, including time-domain indices (e.g., SDNN, RMSSD) and frequency-domain indices (e.g., LF, HF, VLF, and LF/HF ratio). In parallel, overnight polysomnography (PSG) will be performed to determine sleep parameters and OSA severity based on the apnea-hypopnea index (AHI). Clinical data, including demographic characteristics, blood pressure measurements, and relevant laboratory results, will also be collected. Statistical analyses will include descriptive statistics, correlation analysis, and regression modeling to evaluate the relationship between HRV parameters and OSA severity. Machine learning approaches will be applied to develop predictive models for OSA severity using HRV features. Algorithms such as linear regression, support vector machines, random forest, and gradient boosting methods will be evaluated. Model performance will be assessed using metrics including mean squared error (MSE), mean absolute error (MAE), and coefficient of determination (R²), with cross-validation techniques to ensure robustness. In a subsequent phase, participants diagnosed with OSA and receiving standard treatment (e.g., continuous positive airway pressure, CPAP) will be followed longitudinally. Changes in HRV parameters and blood pressure profiles will be analyzed to assess autonomic nervous system recovery and treatment response. This study aims to establish a non-invasive, clinically applicable, and cost-effective strategy for screening and risk stratification of OSA in patients with hypertension, with potential applications in telemedicine and wearable health monitoring systems.

No related papers found yet.

Taiwan