Wrist Pulse Diagnosis Using Complex Network

Pulse signal contains important information about health status and pulse diagnosis has been extensively applied in oriental medicine. In recent years more and more research interests have been given on computerized pulse diagnosis. Pulse feature extraction plays an important role in computerized pulse diagnosis. The most popular pulse feature extraction methods can be grouped into two categories, i.e. time domain feature extraction method and frequency domain feature extraction method. The pulse signal is a pseudo periodic signal while the common feature extraction methods usually assume it is a periodic signal and only a typical period or an averaged period was used in the feature extraction, while the difference between periods was less emphasized. In this paper we use complex network to transform the pulse signal from time domain to network domain and use the statistics parameters which describe the organization of the complex network as the features to characterize the difference between pulse periods. The experiment shows that the complex network features are useful in characterizing the relationship between different pulse periods the diagnosis performance on diabetes are similar with the multi scale sample entropy. By combining complex network features with sample entropy features, higher diagnosis performance can be further obtained.