ECG Signal Medical Assistant Decision System Based On Rough Set

Cardiovascular disease is one of the main chronic diseases that threaten people's life and health in recent years.Patients with cardiovascular disease are often accompanied by arrhythmia symptoms in the early stage.Early detection of arrhythmia symptoms by ECG signal can effectively prevent cardiovascular disease.ECG can reflect the state of heart rate fluctuation.The classification of ECG is an important basis to judge the type of arrhythmia.At present,the methods of ECG classification include neural network,support vector machine,template matching and so on,which have achieved good results in classification success rate.However,the time and space complexity of the existing ECG classification method is high,which is difficult to apply in the miniaturized and portable hardware equipment.Reducing the redundant features of data sets,reducing the resources occupied by classification algorithms,and successfully applying them to practical medical diagnosis are the hot issues in this research field.In this paper,ECG classification and attribute reduction algorithm are studied,and a rough set based ECG medical assistant decision system is proposed1.In this paper,an improved algorithm based on Pan Tompkins algorithm is proposed for the accurate location of ECG feature points.In this algorithm,the first two seconds of ECG signal are used to initialize the adaptive threshold,and then the R wave peak of QRS wave group is located,which realizes the accurate location of ECG feature points and the faster adjustment of the adaptive threshold in the algorithm.Aiming at the problem of using fixed window to intercept ECG signal,which results in the truncation of ECG waveform in a single cardiac cycle,and can't extract the characteristics of ECG signal and waveform at the same time,this paper presents a variable step sliding window method to extract the characteristics of ECG signal,which can retain the characteristics of ECG waveform in a single cardiac cycle and the characteristics of signal in time and frequency domain at the same time,so as to improve the ECG performance characteristic dimension of signal.Experiments show that the algorithm proposed in this paper can locate the feature points of ECG accurately.2.To solve the problem of the effectiveness of the attribute reduction algorithm,this paper uses the information entropy based discretization method to discretize the data set,and uses the attribute reduction algorithm based on the importance of attributes to reduce the attributes of the discrete data set,and verifies the effectiveness of the algorithm through experiments.In the experiment,BP neural network,support vector machine,C4.5 decision tree and cart decision tree are used to classify the UCI data set before and after reduction,and the classification results are compared.Experiments show that the classification accuracy of the data sets before and after reduction is almost the same.Attribute reduction algorithm can effectively remove the redundant attributes in the data set,and improve the efficiency of classification algorithm.3.Aiming at the problem of selecting the best classifier for ECG classification,this paper compares the classification effect of various classifiers on ECG data set,selects SVM as the classifier of medical assistant decision system,and discusses the influence of kernel function and kernel function parameters on the classification performance of SVM.In view of the problem of redundant attributes in ECG data set,which leads to poor classification performance,this paper intercepts ECG signals from MIT-BIH arrhythmia database to form ECG signal sample set,uses the features extracted from ECG signals to build the data set,uses the attribute importance based reduction method to reduce the attributes of the data set,and compares the pre reduction and post reduction of SVM through experiments the classification effect of ECG data set.Experiments show that the classification accuracy of SVMs for ECG data sets before and after reduction is almost the same.Reduction of data sets can improve the training and classification speed of SVMs.The combination of rough sets and traditional classification methods can achieve complementary advantages.4.In this paper,the medical assistant decision system is implemented on the embedded Linux operating system,and the attribute reduction algorithm implemented in this paper is applied to the ECG signal classification of the system.In this paper,we use Python to realize attribute reduction algorithm based on attribute importance,and C + + to realize the basic framework and the construction of each module of the medical assistant decision system.This paper shows the working interface of the medical assistant decision system,which shows the feasibility of ECG signal medical assistant decision system based on rough set.