Research On The Analysis Of The Electrocardiogram Of Patients With Chronic Heart Failure Based On The Support Vector Machine Method

Aims:Electrocardiography was an cheap and noninvasive examination,which was applicable for detecting heart disease.Support vector machine(SVM)was a kind of machine learning,which was popular and advanced.Support vector machine brought new opportunity for analyzing ECG datas in order to identify heart failure.This study aimed to classify ECG datas by support vector machine to identify chronic heart failure with reduced left ventricular ejection fraction,or other cardiac diseases.Methods:A retrospective cohort study was performed.Between the January 2017 and the January 2018,patients who were diagnosed with chronic heart failure with reduced left ventricular ejection fraction(Group A,270 cases)and those(Group B,270 cases)with sinus arrhythmia(SaR)or myocardial infarction(MI),left ventricular high voltage(LVHV)or bundle branch block(BBB),were included in this study.Clinical datas and ECGs were collected.Time domain,frequency domain,nonlieaner domain features were exacted and classifications were performed by analyzing the ECG data features with SVM.Results:Significant differences(P<0.05)were found between Group A and Group B in ECG features,including RRmean、RRSD、TINN、HRVTI、TP、VLF、LF、HF、DET and CD.Accuracy of the SVM could reach 90%in classification of Group A and Group B.Public data of PhysioNet database was also applied to test the efficiency of the SVM model,with the accuracy of 95%.Conclusions:ECG features,including RRmean、RRSD、TINN、HRVTI、TP、VLF、LF、HF、DET and CD,could be used to describe the ECG data features of the chronic heart failure with reduced left ventricular ejection fraction.Support vector machine can establish effective model to classify with patients who diagnosed with chronic heart failure with reduced left ventricular ejection fraction or other cardiac diseases.