Signals analysis of pulse-diagnosis in TCM by the combination of unsupervised learning and supervised learning

Abstract

Objective: To build a viable method to analyze the Pulse-diagnosis signals by the combination of unsupervised learning and supervised learning. Methods: After collecting the pulse waves of patients with chronic obstructive pulmonary disease (COPD) and healthy volunteers, we do pretreatment, parameter extracting basing on harmonic fitting, modeling, and identification by unsupervised learning Principal Component Analysis (PCA) and supervised learning Least-squares Regression (LS) and Least Absolute Shrinkage and Selection Operator (Lasso) with cross-validation step by step for analysis. Result: There is significant difference between COPD patients' pulse waves and the healthy volunteers, and the identification accuracy is about 80%. Features at 3 specific places in radial artery called "youcun" "youguan" "zuochi" in pulse-diagnosis of Traditional Chinese Medicine (TCM) we exacted from the 193 parameters by the 3 methods are especially significant compare with others, and C2youguan is the common one. Conclusion: The method we built basing on the combination of unsupervised learning PCA and supervised learning LS and Lasso is feasible in analyzing the Pulse-diagnosis signals. Furthermore, according to the result of cross-reference by 3 methods and the equation established by Lasso, we can achieve a reliable result by signals of pulse-diagnosis in TCM to identify the healthy volunteers and the patients with COPD. This study might offer some confidence for the realization of computer-aided diagnosis by pulse-diagnosis in TCM, and some important evidence for the scientific of pulse-diagnosis in TCM clinical diagnosis.