Study on the Two-dimensional Sample Entropy of Sleep Apnea Based on the Hilbert-Huang Time-frequency Diagram

Abstract

Sleep apnea (SA) as a common breathing disorder, has been determined to affect human physiological activities and is related to many diseases. Heart rate variability (HRV) analysis as an analysis method of the cardiac autonomic nervous system, is widely used in the study of sleep apnea. The Hilbert Huang Transform (HHT) method is composed of empirical mode decomposition (EMD) and Hilbert spectrum analysis, and is mainly used in nonlinear and non-stationary signal analysis. The two-dimensional sample entropy (SampEn2D) method can effectively analyze the irregularity of the image and evaluate the complexity of the image. We applied SampEn2D to the Hilbert-Huang time-frequency diagram to analyze the complexity of the time-frequency diagram of normal people and patients with sleep apnea. In the study, 60 electrocardiogram recordings were used for analysis, and nonlinearity SampEn2D was calculated. The SampEn2D of sleep apnea patients with different disease severity has significant differences (p<0.05), and the screening accuracy, sensitivity, and specificity reach 90%, 87.5%, and 95%, respectively. The results show that the two-dimensional sample entropy based on the Hilbert-Huang time-frequency diagram can be used to analyze the severity of sleep apnea and SA screening.