Coronary heart disease recognition based on dynamic pulse rate variability

Abstract

Objective: In order to improve the accuracy and real-time of coronary heart disease (CHD) recognition, we propose a new method to analyze the pulse signal with the idea of sliding window iterative. Methods: Firstly, the principle of the feature extraction method(including time domain method, Poincare plot and information entropy) that combined with the idea of sliding window iterative is described. Secondly, The continuous blood pressure signals from the website database PhysioNet are chosen to generate the dynamic pulse rate variability (DPRV) signal as experimental data, and the linear and nonlinear feature is selected for classifying the healthy people and patients with CHD. Finally, the running time and accuracy of the method in this paper are comparaed with other methods. Result: The pulse signal can be online analyzed by this method. The average recognizing accuracy is 97.6%. Conclusion: This methods is entirely feasible. Compared with existing methods, its accuracy and real-time is higher.