Research And Implementation Of Atrial Fibrillation Automatic Recognition Algorithm

Arrhythmia seriously affects physical health,especially atrial fibrillation.Due to the lack of portable detection instruments for atrial fibrillation in life,the detection of atrial fibrillation mainly relies on hospital equipment at present.Now that the prevalence of atrial fibrillation is getting higher and higher.Portable detection equipment can facilitate people to monitor the occurrence of atrial fibrillation at any time in life and achieve the purpose of timely treatment,so it is very meaningful to study a portable identification instrument for atrial fibrillation.This paper studies the automatic recognition algorithm of atrial fibrillation that can be used in portable instruments.The specific content is as follows:Firstly,the preprocessing of Electrocardiograph(ECG)signals and the recognition of R wave were studied.Preprocessing the single-lead ECG signals with a simple integer coefficient filter to make the signal waveform closer to the real heartbeat waveform of the human body.Preprocessing can help to identify the R wave of the signal accurately.Combined with single-lead signal characteristics,the improved algorithm of difference threshold method is given.Simulation experiments show that the improved algorithm improves the detection rate of the R wave of the signal,which lays the foundation for the R wave feature extraction during signal classification.Secondly,the classification algorithm of ECG signals of atrial fibrillation is studied.Based on the characteristics of ECG signals a decision tree algorithm is given to complete the classification and recognition of ECG signals.Three CART decision tree classifiers based on different features are trained to verify the influence of signal feature quality and quantity on decision tree classification.The results show that the feature filtering improves the classification efficiency of the classifier.At the same time,in order to verify the impact of the classification type on the classifier,the two-class and four-class experiments of the CART decision tree on the signal and the four-class experiment of the C4.5 decision tree on the signal were implemented.The experimental results prove that when the CART decision tree is used to divide the signal into four types,higher sensitivity and accuracy can be achieved.Finally,to further improve the accuracy of ECG signal classification algorithm for atrial fibrillation,random forest algorithm and Adaboost algorithm are given.And these two algorithms are implemented to classify and identify ECG signals.Experimental simulation results show that the random forest algorithm has higher sensitivity and accuracy when identifying ECG signals and verify the random forest classification algorithm has a good ability to detect the ECG signal of atrial fibrillation.