Research On Real-time Monitoring Algorithm Of Atrial Fibrillation Based On Intrinsic Recurrence Quantitation Analysis And Wavelet Entropy

The various complications caused by atrial fibrillation continue to threaten human health and bring high cost of treatment to patients.The emergence of mobile wearable devices makes it possible to detect atrial fibrillation in real time,while reducing the workload of doctors to observe and diagnose atrial fibrillation.At present,the detection accuracy of the existing real-time monitoring detection methods for atrial fibrillation is not high enough.In view of the shortage of early-stage atrial fibrillation electrocardiogram signals and the inability to extract effective features,this paper studies the detection of atrial fibrillation in electrocardiogram and proposes a highly sensitive and specific detection method for atrial fibrillation.This paper choose to extract the RR interval as the starting point of the algorithm research.Based on literature analysis,it is found that wavelet transform has the advantage of processing signals without loss of time and frequency resolution,and information entropy represents the average information provided by each symbol.The combination of them can measure the uncertainty of state distribution of dynamic systems and realize the quantification of time series signal information.Recurrence analysis is an effective tool to delineate and quantify the dynamics of complex dynamic systems,e.g.,laminar,divergent or nonlinear transient behaviors.Oftentimes,the effectiveness of recurrence quantification rests on the accurate reconstruction of the state space from univariate time series.This paper presents novel intrinsic recurrence quantification analysis to quantify the recurrence behaviors in complex dynamic systems with short-term observations.As opposed to the traditional recurrence analysis,we delineate and quantify recurrence dynamics in intrinsic scales,which captures not only nonlinear but also nonstationary behaviors in short-term time series.Comprehensive use of intrinsic recursive quantitative analysis and wavelet entropy to identify early-stage atrial fibrillation.Irregular changes of early atrial fibrillation can be detected and frequent ectopic phenomena can be found.It was shown to identify the conditions of early-stage atrial fibrillation with an average accuracy of 88.1% and specificity of 91.5%.The proposed intrinsic recurrence framework can be potentially extended to other nonlinear dynamic methods that are limited in the recording scope.