Coupling analysis for systolic, diastolic and RR interval time series using multivariable fuzzy measure entropy

In this study, we performed a new multivariate fuzzy measure entropy (mvFME) analysis on the recorded multivariate systolic, diastolic and RR interval time series. Twenty healthy young male subjects (24.2±1.9 years) were enrolled. For each subject, both ECG and aortic phonocardiogram (PCG) signals were simultaneously recorded for 5 minutes, under two physiological states respectively: rest and after stair climbing. RR interval time series were constructed from locating QRS complexes in ECG signal by Pan & Tompkins method. Systolic and diastolic time series were constructed from identifying the beginning of the first and second sound in PCG signal by Springer's hidden semi-Markov model segmentation method. The results showed that, compared with rest state, after stair climbing state has significant lower mvFME values for both univariate and multivariate time series analysis (all P<;0.01, except univariate systolic time series with P<;0.05). The mean mvFME values decreased from using univariate to multivariate time series for both rest and after stair climbing states. This study shows physical activity changes the coupling relationship in cardiac interval time series. Meanwhile, coupling between RR and systolic time series reports larger mvFME values than coupling between RR and diastolic time series.