Classification of Wrist Pulse Blood Flow Signal Using Time Warp Edit Distance

The blood flow signals can be used to examine a person’s health status and have been widely used in the study of the clinical diagnosis of cardiovascular diseases. According to the pulse diagnosis theory of traditional chinese, the pathological changes of certain organs could be reflected on the wrist pulse signals. In this paper, we use Doppler ultrasonic device to collect the wrist pulse blood flow signals from patients with pancreatitis (P), duodenal bulb ulcer (DBU), appendicitis (A) and acute appendicitis (AA) as well as healthy persons. After extracting the envelopes of ultrasonic pulse contour, the wrist pulse blood flow signals are pre-processed using wavelet transform. Finally, we adopted a recent time series matching method, time warp edit distance (TWED), on the pre-processed data for classification of wrist pulse blood flow signals. The proposed approach is tested on the wrist blood flow signal dataset, and achieves higher classification accuracy than several classical time series matching approaches, such as Euclidean distance (ED), dynamic time warping (DTW), and edit distance with real penalty (ERP).