Multiscale Distribution Entropy Analysis of Short-Term Heart Rate Variability

Figure 1. Empirical probability density function (ePDF) of white noise with length N=1000 (m=2, r=0.15σ).

Figure 2. Progress of two multiscale processes: (a) coarse-graining; (b) moving-averaging.

Figure 3. Examples of inter-beat (RR) interval time series extracted from the electrocardiogram (ECG) signals of Dataset I: Congestive heart failure (CHF) patient (top panel); Healthy elderly subject (middle panel); Healthy young subject (bottom panel).

Figure 4. Entropy values for synthetic data: (a) multiscale entropy (MSE) value for N = 100; (b) multiscale distribution entropy (MDE) value for N=100; (c) MSE value for N=300; (d) MDE value for N=300; (e) MSE value for N=1000; (f) MDE value for N=1000. Scales between 1 and 20 are used and the value at each scale represents a mean ± standard deviation.

Figure 5. MSE, multiscale permutation entropy (MPE), and MDE results using real data (RR interval) for CHF patients, healthy elderly and healthy young groups: (a) MSE for N=100; (b) MPE for N=100; (c) MDE for N=100; (d) MSE for N=1000; (e) MPE for N=1000; (f) MDE for N=1000. Scale range of 1–20 were used and the value at each scale represents a mean ± standard deviation.

Figure 6. MSE, MPE, and MDE results using real data (RR interval) for other CHF patients and normal sinus rhythm (NSR) subjects: (a) MSE for N=100; (b) MPE for N=100; (c) MDE for N=100; (d) MSE for N=1000; (e) MPE for N=1000; (f) MDE for N=1000. Scale range of 1–20 were used and the value at each scale represents a mean ± standard deviation.