Diagnosis of Sleep Apnea by Evaluating Points Distri?bution in Poincare Plot of RR Intervals

Abstract

Sleep apnea detection using ECG-derived parameters is non-invasive and inexpensive. This article presents novel ECG-derived features to be used in conjunction with existing standard features for improving the detection of sleep apnea. The features presented here were derived using Poincare plots of RR intervals. Global features are based on counting number of points above, below, and on the identity line in Poincare plot. Furthermore, local features are based on point to point variations relative to the identity line (i.e., temporal information in Poincare plot). Performance of features in detection of apnea was evaluated using k-nearest neighbor, self-organizing map, and multilayer perceptron neural network. The accuracy of classifiers on test set was respectively 88.89%, 77.77%, and 100%.