Abnormal ECG Information Identification And Classifcation Decision Tree Based ECG Prrdiction

Heart disease is a high incidence of disease in the population.With the aggravation of aging,automatic analysis and diagnosis technology based on ECG signals will be an important development direction of health care.The prediction and diagnosis of cardiac abnormalities are based on the characteristic points of ECG signals and the information of characteristic bands.It is necessary to ensure the accuracy of the data before extracting the feature information,Filtering the signal is important.Abnormal prediction and diagnosis based on ECG characteristics information needs to collect a large amount of ECG data.The use of machine learning mining algorithms can quickly and effectively diagnose common abnormalities.This paper focuses on ECG signal preprocessing,ECG feature extraction and mining algorithms are studied.The main contents are as follows:1.The job of ECG signal pre-processing is to denoise ECG signals.Second-order filtering of ECG signal in MIT-BIH ECG library is done by using the method of stationary wavelet transform combined with threshold denoising.Residual signal in the baseline drift and electromyography interference are eliminated.In order to overcome the problem of signal distortion caused by using soft threshold function filtering and the pseudo-Gibbs phenomenon caused by the signal discontinuity using the hard-thresholding function,this paper proposes a filtering method using adaptive soft threshold function which use a global adaptive threshold and a soft threshold function to reconstructed a signal.The simulation results show that this method has a good filtering effect on the two kinds of noise,and the performance of the noise and distortion ratio is good.2.ECG signal waveform detection has been studied in feature recognition.The quadratic B-spline wavelet is selected as the wavelet function to decompose the ECG signal into 4 layers.The R-wave modulo extremum pair is detected on the fourth scale,then use adaptive windowing method to achieve R wave detection.Based on R wave,the method can be complied to identify Q,S,P and T waves on the second scale.Finally,identify the starting point of the wave group by the method of quadratic difference threshold.3.In accordance with the rhythm of ECG 12 kinds of common abnormal ECG waveform is divided.According to the identified waveform,the diagnostic criteria for ECG abnormalities are quantified.Then put the Abnormal feature extraction to mysql database.Using Classification Decision Tree algorithm to diagnosis and predict ECG.ECG Anomaly Diagnosis Model is impoved by the use of K-Folds Cross-validation and Random Forest Algorithm.Finally,the performance of the prediction model is verified by using the verification data set.