Bivariate Entropy Analysis of Electrocardiographic RR–QT Time Series

Figures

• 

  Figure 1. RR and QT intervals from a representative subject under resting, cycling, and recovering conditions. Top three panels show the electrocardiogram (ECG) data during resting, cycling, and recovering conditions with R-peaks marked in red circle. A zoomed portion (marked by blue box) is also shown next to each panel on the right-hand side for better visualization. Bottom panels visualize the RR and QT interval time series under each condition.

• 

  Figure 2. Parallel removal of anomalous intervals from RR and QT intervals time series. The time series were from the same subject demonstrated in Figure 1 during cycling. Trends were removed from each of them. An impulse rejection filtering was applied to identify spikes (green circle) that were subsequently replaced with the median value of surrounding five samples (left panel). A moving standard deviation with window size of 100 was performed and for each point two threshold values (blue dashed line) that were above or below three times the corresponding standard deviation from the global mean were used to screen extremes (blue circle) for the second time (middle panel). The points corresponding to those extremes were removed simultaneously from both RR and QT intervals time series (right panel).

• 

  Figure 3. Bivariate entropy analysis of RR–QT intervals time series. (A1–D1) Raw data are shown by dots and results from the same subject are connected using line. (A2–D2) Model estimates for each measure under different conditions are shown as mean (circle) and standard error (error bar).