Research On Collection And Decision Tree Based On Wireless Fidelity Transmission Of Heart Sound Signal By Chai Cheng Long

In the field of clinical medicine,the traditional heart sound auscultation is a commonly used method for the diagnosis of cardiovascular disease,which has subjective judgment.Heart sound signal contains a large number of cardiovascular psychological diseases.A set of electronic heart sound acquisition system is developed and analyzed with decision tree fusion algorithm,which can efficiently distinguish normal heart sound signals from abnormal heart sound signals,which is of great significance for clinical diagnosis.In this paper,a set of electronic heart sound acquisition system based on wireless fidelity technology and LabVIEW storage and real-time display is designed.The pick-up part(chest piece)of the heart sound sensor can be used to collect the heart sound signal,and the built-in 9767 p capacitive electret conversion element is used to convert the acoustic signal into electrical signal,and the weak electrical signal is connected with the designed acquisition circuit through the impedance converter.After the A\D conversion built in the STM32 single chip microcomputer,the ESP8266 module is uploaded to the upper computer in Wi Fi mode to complete the acquisition part of the lower computer.After compiling the LabVIEW program,the collected heart sound signal is digitally filtered and stored in the form of text,and displayed on the host computer in real time.The preprocessing of heart sound signal mainly includes db6 soft threshold wavelet denoising and normalized Shannon energy to extract heart sound envelope,complete the segmentation of heart sound signal,and extract the time gate map of the first heart sound and the second heart sound.Three algorithms are selected to study the classification of heart sound signals,which are logic regression algorithm,GBDT algorithm,logic regression and GBDT fusion algorithm,in which logic regression and GBDT fusion algorithm are decision tree fusion algorithm.EMD decomposition of the extracted heart sound signal can decompose the complex heart sound signal into a limited number of eigenmode functions(IMF),use approximate entropy algorithm to obtain the characteristic value of the heart sound signal,and use 80% of the characteristic value of the heart sound signal as the training set,20% is input as a test set into three algorithms to classify and recognize heart sound signals,and different ROC curves are obtained respectively.Using AUC curve analysis,the accuracy of the three algorithms for heart sound signal recognition is 0.77,0.90,0.95 respectively.According to the standard of AUC curve prediction model classifier,the closer the AUC value is,the higher the accuracy is.It can be concluded that the three algorithms can be used to classify normal and abnormal heart sound signals.Among them,the decision tree fusion algorithm has the highest accuracy for heart sound signal classification,which can reach 0.95,which is of great significance for the study of cardiovascular diseases.