Research On Myocardial Ischemia Detection And Implementation Via Deterministic Learning And Nonlinear Dynamics Analysis

Myocardial ischemia/myocardial infarction is a common cardiovascular disease,which seriously threatens people’s life and health.Electrocardiogram（ECG）is the first choice for the diagnosis of cardiovascular diseases.It has the advantages of non-invasive,simple operation and low price.However,the accuracy of ECG in detecting coronary heart disease/myocardial ischemia is not high（about 60%）.Recently,with the help of deterministic learning theory for accurate dynamic identification of ECG data,a method（CDG）with higher accuracy than ECG has been developed.Based on the non-linear dynamic analysis method,this paper mainly studies ECG feature extraction,classification,recognition and Application Realization Based on CDG.Firstly,the power spectrum method and principal component analysis method are used to analyze the non-linear characteristics of electrocardiodynamics,and the chaotic characteristics in electrocardiodynamics are basically determined.Secondly,four typical non-linear dynamic characteristics are selected:correlation dimension,largest Lyapunov exponent,C₀ complexity and approximate entropy.Quantitative indexes of ECG are extracted.Characteristic distribution of pathological and non-pathological ECG is analyzed in the form of box-plot,and the validity of each feature in detecting myocardial ischemia is preliminarily analyzed.Through the above features,combined with two common classifiers（SVM,KNN）to classify and recognize myocardial ischemia.The results showed that the above non-linear electrocardiodynamic characteristics were clearly distinguished in the classification of myocardial ischemia,and the accuracy of all single features was more than70%.Among them,the best classification effect isC₀ complexity（accuracy 83.79%,sensitivity 84.21%,specificity 81.38%）.Thirdly,considering the above four features,we classify and recognize myocardial ischemia based on multiple features.The results showed that the accuracy,sensitivity and specificity of multi-feature classification were 86.64%,85.82%and 87.45%,which were consistent with the existing clinical results.Aiming at the visualization of multi-feature data,a polygon discriminant map is obtained by combining radar graph representation with SVM optimal classification surface,which provides visual and effective discriminant information for clinical diagnosis of myocardial ischemia.Finally,based on the database of MATLAB and MySQL,an ECG workstation for myocardial ischemia is developed.The workstation can support the functions of local collection,automatic analysis and data management of ECG signals.The program calculation of non-linear ECG characteristics in this paper is realized,which provides quantitative parameters for doctors to diagnose myocardial ischemia for reference in clinical diagnosis,and also for medical staff to manage.Medical record information is easy to provide.