Feasibility of Ultra-Short-Term Analysis of Heart Rate and Systolic Arterial Pressure Variability at Rest and during Stress via Time-Domain and Entropy-Based Measures

Figure 1. (a) Schematic illustration of the experimental protocol, including baseline resting (R1), orthostatic stress (T), second resting (R2) and mental stress (M). Dashed boxes indicate the windows taken into account with regard to short-term (ST, 300 points) analysis. (b) Representative RR and SAP time series, extracted respectively from ECG and BP recordings, which have been investigated through univariate analysis performed after ST (red arrow) and ultra-short-term (UST, 240 to 60 points, blue arrows) time window segmentation.

Figure 2. Boxplot distributions (top subplots) of time-domain indexes, i.e., (a) MEAN, (b) SDNN and (c) RMSSD calculated from RR time series during R1 (light gray) and T (light blue), (.1 panels), and during R2 (dark gray) and M (orange) (.2 panels) phases. Statistical tests: #, p < 0.05, T vs. R1 or M vs. R2; *, p < 0.05, ST vs. UST. Central subplots: Cohen’s d (in absolute value) evaluated between each stress condition and the previous rest phase (i.e., R1-T and R2-M) for all the considered time series lengths. Bottom subplots: squared Pearson correlation coefficients computed between a given UST distribution and the ST reference, with a threshold of r2=0.81 (dotted gray line).

Figure 3. Boxplot distributions (top subplots) of time-domain indexes, i.e., (a) MEAN and (b) STD calculated from SAP time series during R1 (light gray) and T (light blue), (.1 panels), and during R2 (dark gray) and M (orange) (.2 panels) phases. Statistical tests: #, p < 0.05, R1 vs. T and R2 vs. M; *, p < 0.05, ST vs. UST. Statistical tests: #, p < 0.05, T vs. R1 or M vs. R2; *, p < 0.05, ST vs. UST. Central subplots: Cohen’s d (in absolute value) evaluated between each stress condition and the previous rest phase (i.e., R1-T and R2-M) for all the considered time series lengths. Bottom subplots: squared Pearson correlation coefficients computed between a given UST distribution and the ST reference, with a threshold of r2=0.81 (dotted gray line).

Figure 4. Results of information domain analysis on RR time series. Boxplot distributions (top subplots) of ST and UST indices of (a) SE, (b) DE and (c) CE calculated using both lin (.1 and .2) and knn (.3 and .4) estimators during R1 (light gray) and T (light blue) (.1 and .3), and during R2 (dark gray) and T (orange) (.2 and .4) phases. Statistical tests: #, p < 0.05, T vs. R1 or M vs. R2; *, p < 0.05 ST vs. UST. Central subplots: Cohen’s d (in absolute value) evaluated between each stress condition and the previous rest phase (i.e., R1-T and R2-M) for all the considered time series lengths. Bottom subplots: squared Pearson correlation coefficients computed between a given UST distribution and the ST reference, with a threshold of r2=0.81 (dotted gray line).

Figure 5. Results of information domain analysis on SAP time series. Boxplot distributions (top subplots) of ST and UST indices of (a) SE, (b) DE, and (c) CE calculated using both lin (.1 and .2) and knn (.3 and .4) estimators during R1 (light gray) and T (light blue) (.1 and .3), and during R2 (dark gray) and T (orange) (.2 and .4) phases. Statistical tests: #, p < 0.05, T vs. R1 or M vs. R2; *, p < 0.05 ST vs. UST. Central subplots: Cohen’s d (in absolute value) evaluated between each stress condition and the previous rest phase (i.e., R1-T and R2-M) for all the considered time series lengths. Bottom subplots: squared Pearson correlation coefficients computed between a given UST distribution and the ST reference, with a threshold of r2=0.81 (dotted gray line).