LAUSR

Objective: Background. Timely verification and treatment of sleep disorders are crucial for the prevention and control of cardiovascular diseases, above all hypertension. However, specialized sleep diagnostics is accessible only for a minority of patients. The development of various diagnostic approaches and screening tools appears to be helpful. In particular, routinely used long-term electrocardiogram (ECG) and blood pressure (BP) monitoring is promising for simultaneous assessment of both cardiovascular and sleep parameters. The aim of our study was to develop an objective approach for the identification of sleep structure and sleep-wake cycle based on the cardiorespiratory and activity parameters recorded during the long-term ECG and BP monitoring. Design and method: We recorded simultaneously 24-h cardiovascular (12-lead ECG and BP; three axis accelerometer fixed on the right V intercostal space; impedance pneumography) monitoring (Kardiotekhnika, Inkart, Russia) and nocturnal PSG (Embla N7000, Natus, USA) in 23 subjects (aged 17–75 years, 13 males). Based on PSG analysis, hypnograms (sleep structure) were verified by an experienced specialist. From ECG/BP monitoring the data from accelerometer, ECG and respiratory pattern were obtained. Based on diurnal activity sleep and wake periods were identified (modified Koele and Kripke's algorithm). Based on the analysis of heart rate variability and respiration variability a specified classification of sleep stages was developed (Fig.). The accuracy, sensitivity and specificity of the developed algorithms were evaluated. Figure. No caption available. Results: Our approach based on actigraphy analysis enabled classification of sleep and wakefulness with specificity of 80% and sensitivity of 86%. The combined analysis of both heart rate variability and respiratory signals allowed a 3-stage sleep classification (light, deep and rapid-eye-movement sleep) with the accuracy of 71.4% and comparatively high inter-rater agreement (Cohen's kappa 0.58 ± 0.16). Sleep efficiency error was 6.7 ± 6.6%, total sleep time error - 33.2 ± 45.3 min, sleep onset latency error - 22.3 ± 35.8 min. Conclusions: The implementation of the proposed combined analysis of activity, heart rate variability and respiration pattern in the ECG/BP monitoring systems is a promising alternative to specialized sleep diagnostic studies to be used in routine cardiovascular medicine practice.