The Research On J Wave Diagnosis Based On The Combination Of Linear And Nonlinear Features

Clinical analysis shows that there is a certain intrinsic connection between the generation of J wave signals and the degree of heart disease.The appearance of J wave indicates that a series of cardiovascular diseases such as malignant arrhythmia,myocardial infarction,and sudden cardiac death are very likely to occur.In order to reduce the diagnosis pressure of doctors and improve the recognition efficiency of J wave related diseases,this paper extracts the features of ECG signals and adopts machine learning methods by using a computer-aided diagnosis model to diagnose J wave,thus achieving efficient identification of J wave.Through digital processing,this paper converts paper ECG signals obtained from hospitals into digitally stored data,and completes the establishment of J wave database.Three types of diagnostic J wave models were proposed for comparison:Firstly,a J wave diagnosis model based on linear features was proposed.The three methods of wavelet transform,empirical mode decomposition(EMD)and extreme-point symmetric mode decomposition(ESMD)were used to extract the normal ECG signals and the two kinds of ECG signals with J wave ECG respectively.The extracted features are reduced in dimension by principal component analysis.Then neural network,k-Nearest Neighbor(KNN),and support vector machine(SVM)are used for training and testing respectively.The results show that the J wave diagnostic model has the best performance when ESMD and SVM are used,and the accuracy is 90.03%.Then,a J wave diagnosis model based on nonlinear features was proposed.The nonlinear features of ECG signals were extracted by high-order cumulants(HOC)method and chaotic analysis method.The dimensionality was reduced by independent component analysis,and the processed data were input to Neural network,KNN,and SVM were used to classify and identify J wave diagnosis models.The results show that when using all nonlinear features and SVM,the best effect is achieved,the accuracy rate is 90.86%;The J wave diagnosis model proposed in the end is to fuse the above two models,combine linear features and nonlinear features,and optimize the parameters of the SVM.By contrast,when using the SVM by the artificial bee colony algorithm optimized and fusion features,the highest diagnostic accuracy is 97.85%,which can effectively identify the J wave.In addition,this paper further discusses the ECG big data model and proposes the SVM based on Hadoop platform,the accuracy rate reaches 98.28%.