Pulse Manifestation Analysis Of Human Body Based On Multi-domain Feature Selection

The pulse diagnosis of traditional Chinese medicine(TCM)is an indispensable part of traditional Chinese medicine.The pulse diagnosis of traditional Chinese medicine can be used for disease diagnosis and preventive treatment of disease.Pulse manifestation information can reflect the physiological and pathological information of the internal organs of the human body,and this information is obtained through the doctor's finger feeling and clinical experience.The representation of the pulse manifestation information lacks a unified standard,and it is difficult to objectify the pulse manifestation.Realizing the Objectification of TCM pulse manifestation has the practical need to propel and spread research in the field of TCM.In this paper,the pulse manifestation information of three different populations such as healthy college students,pregnant women,and middle-aged and elderly patients are taken as the research object,to study the features of the pulse manifestation of different populations,to analyze the features by extracting the multi-domain features of the pulse manifestation signal,and to combine the three commonly used features selection algorithm is to select multi-domain typical features that can reflect the pulse manifestation difference of college students with different constitution,SVM classifier is used to realize college students' constitution recognition.The work that has been completed includes:(1)Firstly,pulse manifestation database of healthy college students,pregnant women,middle-aged and old people with chronic diseases was established,among which 307 cases were healthy college students,33 cases were pregnant women,and 75 cases were middle-aged and old people with chronic diseases,which provided data support for subsequent pulse manifestation analysis of different groups.(2)Secondly,multi-domain feature extraction and differential analysis were carried out on pulse manifestation signal.After preprocessing,the multi-domain features of pulse manifestation signals were extracted,and 132 features in time domain,frequency domain and time domain were obtained.The correlation and difference between the pulse manifestation features of the three parts of chi,guan and cun,and the difference between the left and right pulse manifestation features were analyzed.The difference of pulse manifestation features with different genders and different constitution of the healthy college students,the difference of pulse manifestation features between pregnant women and non-pregnant female college students,and the difference of pulse manifestation features between middle-aged and old patients with chronic diseases and healthy college students were analyzed.The results show that there are differences in the distribution of some pulse manifestation features of different types of people.(3)Finally,combining with the classification effect of multi-domain features and single domain features,this paper completes the body constitution recognition of college students based on multi-domain pulse manifestation features based on three feature selection methods,namely,Gini index,Fisher Score,t-test.The results show that the multi-domain feature is helpful to improve the recognition effect and the use of multiple feature selection methods is helpful to obtain the typical correlation features which can stably characterize the physical differences.On this basis,based on these multi-domain typical pulse manifestation features,the SVM classifier was used to carry out the identification of different constitution of college students.The results show that the application of multi-domain pulse manifestation features to constitution recognition is effective.