Complex Analysis of Heart Rate on Obstructive Sleep Apnea using Fuzzy Approximate Entropy

Abstract

Obstructive Sleep Apnea (OSA) is an easily overlooked disease related to abnormal autonomic nerve system (ANS), which can be measured using Heart rate variability (HRV). A classical method like time-domain analyses and frequency-domain analyses are linear methods. They can't analyse the nonlinear autonomic nerve system correctly, and the ability to measure complexity is weak. Therefore, Fuzzy Approximate Entropy is introduced as a nonlinear method to extract information features from HRV signals. This paper analyzed 30 PPG recordings (15 OSA, 15 normal), the length of these recordings are 6-7 hours and were divided into 5-minutes time slices. Compared with the classical method, the Fuzzy Approximate Entropy method shown a significant difference (p<0.01) and the highest accuracy of 76.7%. When combining with STD and LF/HF power ratio, the classification result reached an accuracy of 86.75%, sensitivity of 93.3% and specificity of 80%. This study showed that OSA patients have a lower level of confusion in HRV signals, which means increased repetition patterns between heartbeats. Therefore, FuzzyEn can be used as a new indicator in OSA screening and prefect the classification method.