Research On Discriminant Algorithm Of Sleep Apnea Syndrome Based On Electrocardiographic Signal

Sleep Apnea Syndrome(SAS)is a common sleep disorder that severely affects people's sleep quality and health.Polysomnography(PSG)is a commonly used clinical method to detect sleep apnea syndrome,but it has the disadvantages of uncomfortable and expensive detection.The widespread promotion of electrocardiographs and portable medical devices in the modern era has made the collection of ECG signals increasingly popular and convenient.The literature has shown that ECG signal and respiratory signal have a high correlation with sleep apnea syndrome.The time domain and frequency domain features of respiratory signals can more directly reflect sleep apnea syndrome.Therefore,single-channel ECG characteristics are studied.The acquisition of respiratory signals from singleconductor electrocardiographic signals is important for the diagnosis of sleep apnea syndrome.Therefore,this dissertation focuses on the study of the algorithms for acquiring respiratory signals based on ECG signals and the judgment of sleep apnea,in order to improve the accuracy of the judgment results.In the research of extracting the characteristics of respiratory signal,this paper uses an algorithm based on Indepent Component Analysis(ICA)algorithm to get the respiratory signal by blind source separation.Firstly,the original ECG signal is detected by differential threshold method and the error is corrected.Then the QRS matrix is constructed by window,and the principal component analysis(ICA)is used to center and whiten the QRS matrix.Remove the correlation between data and reduce the data dimension.Six groups of data with high contribution rate are selected as the objects of ICA processing.Six groups of related signals are finally obtained after interpolation.The window-in-window power ratio is calculated through windowing,and the largest value of the ratio is selected as the final extracted respiratory signal characteristics,and the coherence is calculated.The respiratory signal extracted by the ICA is highly coherent with the original respiratory signal.This is used for the next step in the SAS judgment.In the study of classification judgment of sleep apnea syndrome,the power spectrum characteristics of respiratory signals obtained by ICA method were selected for classification.At the same time,the time domain features of HRV and the power spectrum characteristics of respiratory signals obtained by PCA method were selected.As a comparison group.In the support vector machine classification,four kinds of kernel functions are selected: Linear Kernel,Polynomial Kernel,Sigmoid Kernel,and Radial Basis Function(RBF).Finally,when the power spectrum characteristics of respiratory signals were extracted by ICA method and the support vector machine was classified by radial basis kernel function,the best effect was achieved,achieving 91.29% accuracy,89.84% sensitivity,and 91.75% F value.This article selected ECG and respiratory data from PhysioNet's Apnea database and compared the database's “Normal Breathing”(N,Normal)or “Disordered Breathing”(A,Apnea)markers as the final test results.The experimental results show that the algorithm used in this paper has a good effect in improving the accuracy of sleep apnea.