Research On Mobile ECG Diagnostic System

With the continuous improvement of people’s living standards,people’s "three highs" problems are widespread,leading to a rise in cardiovascular disease mortality year by year,and cardiovascular disease has become the main killer of human life and health.In recent years,the number of deaths due to myocardial infarction in cardiovascular diseases has risen sharply.Due to its acute complications,the disease has the characteristics of rapid progress and high mortality.For myocardial infarction,early detection and early treatment should be performed for real-time monitoring.With the development of wireless Internet of Things technology,many wearable ECG detection devices are gradually available,and the massive mobile ECG data brought by it has become a major challenge in the field of automatic analysis of ECG signals.In this paper,from the perspective of processing massive mobile ECG data,ECG signal preprocessing,feature extraction and automatic diagnosis of myocardial infarction in ECG signal analysis are studied,and the characteristics of myocardial infarction ECG signals are explored from various fields.The diagnostic model of myocardial infarction provides a certain reference and has important practical significance for the prevention of cardiovascular disease.The main work of this paper is as follows:(1)ECG signal is a millivolt-level WeChat bioelectric signal,which is very susceptible to interference.It deeply analyzes the noise component and spectral structure of ECG signals,studies the advantages and disadvantages of different threshold functions of wavelet denoising,and improves the weighted threshold shrinkage.Denoising method to better preserve the details of the ECG signal waveform.The commonly used waveform detection algorithm is discussed.Based on the traditional threshold detection and wavelet detection,an ECG waveform detection method based on wavelet method and threshold method is designed.The corresponding scale signal is enhanced by differential means and the adaptive threshold update is adopted.,refractory period and backtracking compensation to achieve accurate positioning of ECG waveforms.(2)The study looked for multi-domain features for the automatic diagnosis ofmyocardial infarction,and analyzed and extracted features from time domain,RR Lorenz scatter plot and energy distribution.In the time domain,the waveform detection algorithm combined with the wavelet method and the threshold method is used to detect the ECG waveform and extract the corresponding time domain features.RR Lorenz scatter plot,by analyzing the normal ECG signal and the RR Lorenz scatter plot of the myocardial infarction,the feature information of the scatter plot is extracted,and the heart reflected by the scatter plot is further excavated.System message.In terms of energy distribution,by comparing and analyzing the power spectrum estimation of normal ECG and myocardial infarction,combined with spectrum analysis and clinical electrocardiography,energy characteristics were extracted according to the change of waveform energy after myocardial infarction.(3)Using support vector machine(SVM)as the classifier of myocardial infarction ECG signal,the ECG signals of PTB,CRIS and NSR2 databases were selected as experimental data,and the three sets of characteristics were respectively introduced into SVM.The features extracted by RR Lorenz scatter plot and the features extracted by power spectrum estimation have ideal classification results,which can effectively identify myocardial infarction and normal ECG.