Research On TCM Pulse Factor Classifiers And Asthma Pulse Classification Based On Signal Processing

Pulse diagnosis is an important part of Traditional Chinese Medicine(TCM)diagnostic methods,and its objectification research is one of the key aspects of inheriting and developing TCM.In traditional TCM pulse diagnosis,physicians perceive the body's pulse pattern and identify the disease by distinguishing multiple pulse factors such as pulse rate and pulse position.Most of the current pulse diagnosis objectification researches about pulse waveform signal processing cannot correspond with TCM clinical pulse diagnosis theory,they are not only difficult to reproduce the personalized diagnostic method of pulse diagnosis,but also fail to make the best of human physiological and pathological information embedded in the pulse waveform signal.Based on TCM pulse diagnosis theory,this thesis applied a variety of signal processing and machine learning methods to research the complete pulse waveform signal analysis process including preprocessing,feature extraction and evaluation,pulse classification and recognition.Firstly,methods of pulse waveform signal preprocessing,feature extraction and detecting characteristic points for atypical pulse waveform were proposed.Then a method of extracting respiratory information from pulse waveform signal was proposed,and pulse rate classifier with personalized judgment criteria based on this method was designed and verified;pulse position classifier was designed by processing pulse waveform signals under multiple sampling pressures;pulse shape and pulse trend classifiers were trained through features selection and application of Xgboost,Support Vector Machine algorithms respectively.Finally,this thesis discussed asthma pulse classification in two new perspectives:pulse waveform feature importance assessment and respiratory waveform feature extraction.To verify the feasibility of four pulse factor classifiers in clinical pulse diagnosis,the pulse factor classifiers were applied on the asthma pulse data set,and the classification results were analyzed using Fuzzy C-means Clustering Algorithm(FCM),The results of cluster analysis showed that the pulse clusters of typical asthma patients and healthy people were in accordance with TCM pulse diagnosis theory,indicating that the pulse factor classifiers proposed in this thesis could assist clinical pulse diagnosis to a certain extent.