Session S22.2

Characterization of Heart Rate Variability Loss with Aging and Heart Failure Using Sample Entropy

R Goya-Esteban*, JP Marques de Sá, O Barquero-Pérez, JL Rojo-Álvarez

Instituto de Engenharia Biomédica
Porto, Portugal

Entropy based measures have been widely used for quantifying the Heart Rate Variability (HRV), with the hypothesis that decreasing entropy points to a perturbation of the complex physiological mechanisms or disease. However, higher entropy values have been reported for some pathologies than for healthy subjects, which could be due to the use of a threshold value r tuned relative to the signal standard deviation.
Given that the free parameter tuning in entropy methods is crucial for their performance, we hypothesize that an adequate selection of the threshold value r will yield a better quantification of HRV.
The SampEn statistic was used to assess the variability of the RR-interval signals from 24-hour Holter monitoring in healthy subjects (72 recordings, 20-78 years-old) and in congestive heart failure (CHF) subjects (44 recordings, 22-79 years old). In order to discriminate between healthy and pathologic subjects, we first studied the influence of the free parameters. We then split the recordings into different groups according to the age of the subjects to analyze the evolution of the statistic with the age.
The use of a fixed threshold value r for all the recordings (obtained as a percentage of the mean standard deviation of whole recordings), instead of its conventional setting (as a percentage of the standard deviation of each data series), produced better discrimination between healthy and CHF subjects and higher values for the formers in agreement with the idea of loss of HRV in pathological conditions, (1.00±0.23 vs. 0.51±0.26, p<0.001) for fixed r and (1.09±0.27 vs 1.37±0.36, p<0.001) for variable r. We also found that, with fixed r, it was possible to quantify the loss of HRV due to the aging in healthy subjects (20-30 yrs, 0.91±0.29; 31-40 yrs, 0.86±0.21; 41-50 yrs, 0.76±0.24; 51-60 yrs, 0.67±0.16; 61-70 yrs, 0.61±0.15; 71-80 yrs, 0.54±0.20), which was less clearly present with variable r, since it did not show a fixed increasing or decreasing tendency. A regression line was fitted, for healthy subjects, a variation of 0.0085 units per year, and a determination coefficient 0.36 (p < 0.001) was obtained for fixed r. For CHF subjects, no significant correlation was found between the age and the variation of SampEn value.
By using a fixed threshold value r in the SampEn algorithm, higher discrimination between healthy and CHF subjects was achieved, and a usable aging curve of HRV loss could be built for healthy subjects.

(Abstract Control Number: 162)