ECG Detection And Identification Of Common ECG Abnormalities In The Elderly

With the deepening of aging in China,cardiovascular disease has seriously affected people’s quality of life and social stability.At present,the research on health monitoring and disease diagnosis of the elderly has received extensive attention and recognition,and its research focus is the analysis and diagnosis of abnormal ECG signals.At present,the research on abnormal ECG recognition mainly focuses on the classification of common arrhythmia.There are still some shortcomings such as the low accuracy of ECG signal recognition and the long training time of the classifier.In this paper,the classification of common abnormal ECG signals in the elderly is deeply studied.The classification accuracy is improved by modifying the process of noise and signal feature extraction,and the accurate recognition of seven types of common ECG signals in the elderly is realized.The main contents of this paper are as follows:1.The characteristics of common interference of ECG signal are analyzed and the preprocessing of ECG signal is completed with denoising algorithm.Compared with other literatures,this paper deals with the three main kinds of noise,even if the median filter and the improved LMS filter are used to remove the baseline drift and power frequency interference.Wavelet transform and sparse decomposition are used to study the most influential EMG noise,and artificial bee swarm algorithm is used to improve the matching speed of atoms in sparse decomposition.The experimental results show that the preprocessing method designed in this paper can filter the common interference on the premise of ensuring that the ECG information is not lost and the waveform is not distorted,and the denoising effect is better.2.Focus on ECG feature wave recognition and location and feature extraction.Firstly,a new method of differential threshold combined with phase space reconstruction technology is proposed in this paper,which not only realizes the accurate location of R wave(the false detection rate is reduced by half),but also shortens the system time by nearly 5 times.Then the detection results of the starting and ending point of the characteristic wave are modified,that is,the forward difference method is used to find the starting and ending point of the stable change as the modified starting and ending point.This method effectively avoids the interference of residual noise to feature extraction.Considering that ST segment has important clinical significance and is closely related to myocardial infarction and other diseases,curve fitting and other methods are used to analyze it,and finally the recognition of nine forms of ST segment is realized.3.A classifier for common ECG abnormal signal recognition in the elderly is designed.In this paper,support vector machine(SVM),which has great advantages in dealing with small samples and nonlinear data,is selected as the basic model for the design of classifiers.In order to further optimize the performance of the system,firstly,the feature parameters are simplified by data dimension reduction to shorten the training time,and then the parameters of the classifier are adjusted by using various parameter optimization algorithms.Finally,the design scheme of the KPCA+GA+SVM classifier is determined.The ECG data used in this paper are all from the international standard ECG database,and the samples used for classification training are also extracted according to the expertundefineds comment file,so the experimental results are very reliable and persuasive.Compared with the traditional SVM classifiers,the average accuracy of the ECG waveform classifiers designed in this paper is 98.79%,and the training time is shortened by 11.46%.