Signatures of the autonomic nervous system and the heart’s pacemaker cells in canine electrocardiograms and their applications to humans

Figures

• 

  Figure 1. Canine beat intervals in the basal (BSL) and autonomic nervous blockade (ABK) states. Representative examples from one canine, showing visually reduced variability after ABK: (a) beat intervals; (b) beat interval distribution; (c) Poincaré plot; (d) normalized power spectral density, showing also the elimination of the ANS-related peaks (baroreceptor, 0.15 Hz and respiratory, 0.3 Hz) after ABK. Averaged values of: (e) normalized power in the very low frequency (VLF), low frequency (LF) and high frequency (HF) bands, showing a roughly twofold increase in normalized VLF power after ABK; (f) histogram of normalized power, showing the shift of spectral power density towards the VLF band. Vertical bars represent standard error. Figures rendered with Matlab41.

• 

  Figure 2. Fractal and complexity metrics of beat intervals in basal (BSL) and autonomic nervous blockade (ABK) states of the canine data, compared to ideal white, pink and Brownian noise processes. (a) Average multiscale entropy (MSE); (b) MSE of ideal noise, showing a monotonically decrease for while noise, high complexity on all scales for pink noise and a monotonic increase for Brownian noise; (c) power spectral density (PSD) of the VLF band, with average slope (β); (d) PSD of ideal noise showing a slope of 0, −1 and −2 for white, pink and Brownian noise respectively; (e) detrended fluctuation analysis (DFA); (f) DFA of ideal noise showing a slope of 0.5, 1 and 1.5 for white, pink and Brownian noise respectively. Vertical bars represent standard error. Figures rendered with Matlab41.

• 

  Figure 3. ANS and SAN signatures in canine basal ECG. (a) log power spectral density (PSD) of a generated signal from our model of the ANS contribution to beat intervals. Averaged basal (BSL), autonomic nervous blockade (ABK), and ABK with added samples from ANS model (ABK+ANS), showing dynamics shifted towards the BSL state: (b) multiscale entropy (MSE) and (c) detrended fluctuation analysis (DFA). BSL, ABK, and BSL after application of exponential filter (BSL + EPF), showing dynamics shifted towards the ABK state: (d) averaged multiscale entropy (MSE); (e) averaged detrended fluctuation analysis (DFA); (f) representative Poincaré plot. Vertical bars represent standard error. Figures rendered with Matlab41.

• 

  Figure 4. Human beat intervals in basal (BSL) and autonomic nervous blockade (ABK) states of the Ditor dataset. Representative examples of: (a) beat intervals; (b) beat interval distributions; (c) power spectral density (PSD), showing mostly-eliminated ANS-related peaks (baroreceptor, 0.1 Hz and respiratory, 0.25 Hz) after ABK. Averaged values of: (d) histogram of normalized PSD; (e) multiscale entropy (MSE); (f) detrended fluctuation analysis (DFA). Figures rendered with Matlab41.

• 

  Figure 5. Human beat intervals of healthy (NSR), atrial fibrillation (AF), congestive heart failure (CHF), and aged individuals (AGING) from the PhysioNet datasets. (a) beat interval distributions; (b) histogram of normalized power; (c) multiscale entropy (MSE), showing the characteristic high-entropy at all scales in the NSR, a monotonic decrease indicating no SAN regulation in AF, a reduction with monotonic increase indicating impaired ANS and SAN regulation in CHF and slightly reduced complexity in AGING; (d) detrended fluctuation analysis (DFA), showing slopes in agreement with the MSE analysis. Vertical bars represent standard error. Figures rendered with Matlab41.

• 

  Figure 6. Simulation of a coupled-oscillator model fit to canine beat intervals, under different values of the coupling coefficient. (a) Schematic illustration of the model and its parameters; (b) beat intervals generated by the model with different coupling coefficient (K) values, compared to original canine intervals (RR); (c) beat interval distributions; (d) normalized power spectral density (PSD), showing increased VLF power with higher coupling; (e) multiscale entropy (MSE), showing generally higher values as coupling increases. The model parameters were fitted to canine beat intervals after autonomic blockade, and afterwards only the parameter K was changed by a constant factor to generate the plots. Figures rendered with Matlab41.