Enhanced EMD multiscale poincaré plots (EEMP) of heartbeat time series for visual analysis

Abstract

Multiscale science is an emerging scientific field, and the multiscale computing methods, such as coarse-graining and empirical mode decomposition (EMD), have received increasing attention. Heartbeat Time Series are rich in multiscale information, but the traditional analysis methods only consider them on the single scale, which is not conducive to a comprehensive understanding of the dynamics of the system. Meanwhile, poincaré plot is widely used in the analysis of heart rate variability (HRV) which quantifies the variability of the series, but without probability distribution. To facilitate multiscale visual analysis of RR intervals series, we introduce enhanced EMD multiscale poincaré plot (EEMP). We first employed EMD to create a family of intrinsic mode function (IMF), each of which represents the characteristic time scale embedded in the series. Next, multiscale poincaré plots are constructed for each scale. Finally, to display probability distribution intuitively, color labels are added to each point using kernel density estimation to exhibit the normalized frequency. We illustrated the EEMP approach on RR intervals time series from normal subjects comparing with atrial fibrillation (AF) subjects and congestive heart failure syndrome (CHF) subjects to analyze the dynamics of different physiology states visually. In the future, this generalized approach may be used well in other types of time series.