Obstructive Sleep Apnea Heart Rate Variability Analysis using Gramian Angular Field images and Two-dimensional Sample Entropy

Abstract

Obstructive Sleep Apnea (OSA) is a sleep-breathing disorder accompanied by multiple complications, and often associates with autonomic dysfunction. Sample entropy based on Gramian Angular Summation Field image (CSpEn2D) for OSA autonomic nervous system (ANS) study and analysis. We used 60 ECG records from the Physionet database. Low frequency to high frequency power (LF/HF) ratio could not distinguish normal OSA group from moderate OSA group, while CSpEn2D could significantly distinguish normal OSA group, mild-moderate OSA group and severe OSA group (P < 0.05). In terms of disease screening, the accuracy of CSpEn2D was 90.0% higher than that of LF/HF. At the same time, the CSpEn2D and apnea hypoventilation index (AHI) correlation significantly stronger (|R| = 0.727, p = 0). Hence, the CSpEn2D takes in a certain degree of clinical application prospects, and It is an effective indicator of OSA single feature screening.