Research On Portable Three-channel Pulse Measurement System And Pulse Signal Processing

Pulse diagnosis is one of the four diagnostic methods of Traditional Chinese Medicine(TCM),and it plays an important role in the process of TCM syndrome differentiation.In the traditional pulse diagnosis process,TCM physicians judge pulse types based on subjective feelings and long-term accumulated experience,and different doctors may give different diagnoses.In order to make the pulse diagnosis more objective and standardized,modern technology should be used to study the pulse acquisition system and process and analyze the pulse signal.This study designed a portable three-channel pulse measurement device,which used pressure sensors to collect pulse at cun,guan and chi position.Pumps and air bags were used to realize the function of pressurization to simulate the way of taking pulses in TCM.The signal processing circuit was designed,and the pulse signal and pulse pressure signal were obtained by filtering and amplifying the sensor signal.A combined acquisition bracelet was designed to support single and three-channel pulse acquisition.The single and three-channel pulse acquisition experiments were completed using the device.The proposed system is small in size,light in weight,and easy to carry.The characteristics of various combinations can meet various needs of home medical treatment and experimental research.In this paper,pulse pressure signal was processed and the automatic addition of the missing points in atypical pulse signal was realized.Random forests was used to study the importance of pulse features,which can provide guiding suggestions for feature extraction to prevent feature loss and difficulties in identification.The relationship between research characteristics and TCM theory were studied.Feature dimension was reduced based on the feature importance research results.Ssupport vector machine and three ensemble learning methods were used in pulse classification.The research shows that the random forest method has good performance in classification and can be applied to the pulse types recognition research.