entropy

Article
Effects of Cardiac Resynchronization Therapy on
Cardio-Respiratory Coupling
Nikola N. Radovanović 1,2, *, Siniša U. Pavlović 1,2 , Goran Milašinović 1,2 and Mirjana M. Platiša 3

1 Pacemaker Center, University Clinical Center of Serbia, 11000 Belgrade, Serbia; [email protected] (S.U.P.);
[email protected] (G.M.)
2 Faculty of Medicine, University of Belgrade, 11000 Belgrade, Serbia
3 Institute of Biophysics, Faculty of Medicine, University of Belgrade, 11129 Belgrade, Serbia;
[email protected]
* Correspondence: [email protected]; Tel.: +381-11-366-3690; Fax: +381-11-362-9095

Abstract: In this study, the effect of cardiac resynchronization therapy (CRT) on the relationship
between the cardiovascular and respiratory systems in heart failure subjects was examined for the
first time. We hypothesized that alterations in cardio-respiratory interactions, after CRT implantation,
quantified by signal complexity, could be a marker of a favorable CRT response. Sample entropy
and scaling exponents were calculated from synchronously recorded cardiac and respiratory signals
20 min in duration, collected in 47 heart failure patients at rest, before and 9 months after CRT
implantation. Further, cross-sample entropy between these signals was calculated. After CRT,
all patients had lower heart rate and CRT responders had reduced breathing frequency. Results
revealed that higher cardiac rhythm complexity in CRT non-responders was associated with weak





 correlations of cardiac rhythm at baseline measurement over long scales and over short scales at
follow-up recording. Unlike CRT responders, in non-responders, a significant difference in respiratory
Citation: Radovanović, N.N.;
Pavlović, S.U.; Milašinović, G.;
rhythm complexity between measurements could be consequence of divergent changes in correlation
Platiša, M.M. Effects of Cardiac properties of the respiratory signal over short and long scales. Asynchrony between cardiac and
Resynchronization Therapy on respiratory rhythm increased significantly in CRT non-responders during follow-up. Quantification
Cardio-Respiratory Coupling. of complexity and synchrony between cardiac and respiratory signals shows significant associations
Entropy 2021, 23, 1126. https:// between CRT success and stability of cardio-respiratory coupling.
doi.org/10.3390/e23091126
Keywords: heart failure; cardiac resynchronization therapy; cardiopulmonary coupling; heart
Academic Editor: José A. Tenreiro rhythm; respiratory rhythm; responders; non-responders; sample entropy; cross-sample entropy;
Machado detrended fluctuation analysis

Received: 29 June 2021

Accepted: 8 August 2021
Published: 30 August 2021
1. Introduction
Publisher’s Note: MDPI stays neutral
Heart failure is an increasingly prevalent disease accompanied by significant mor-
with regard to jurisdictional claims in
bidity and mortality [1]. It is predicted that hospital admissions due to heart failure and
published maps and institutional affil- the costs of treating these patients will continue to rise in the next decade, [2]. Cardiac
iations. resynchronization therapy (CRT) is a well-established therapy for appropriately selected
symptomatic patients with heart failure and reduced left ventricular ejection fraction [3]. In
most recipients, it slows disease progression, alleviates symptoms, reduces emergency ser-
vice visits, hospitalizations, and prolongs life [4]. The benefits of cardiac resynchronization
Copyright: © 2021 by the authors.
therapy have been explained primarily by the increase in the electromechanical synchrony
Licensee MDPI, Basel, Switzerland.
of the ventricles, i.e., by the decrease of conduction delay [5]. However, today we know
This article is an open access article
that CRT implantation leads not only to mechanical and electrical changes but also to
distributed under the terms and changes in the level of neurohormonal activity [6–8]. Imbalance of the autonomic nervous
conditions of the Creative Commons system (ANS) is part of the basic pathophysiological characteristic of heart failure [9]. So
Attribution (CC BY) license (https:// far, only few studies have examined the ANS modifications that occur in patients after
creativecommons.org/licenses/by/ CRT implantation. We know for sure that in all CRT patients, in the first hours after device
4.0/). implantation, an improvement in cardiac neuroadrenergic function is registered, confirmed

Entropy 2021, 23, 1126. https://doi.org/10.3390/e23091126 https://www.mdpi.com/journal/entropy

Entropy 2021, 23, 1126 2 of 15

by increased levels of presynaptic catecholamine uptake, which is a direct acute hemody-

namic effect of left and right ventricular synchronization [10,11]. However, the results of
chronic follow-up in a small number of patients show that the improvement in sympa-
thetic function, i.e., the increase in catecholamine uptake, is observed exclusively in CRT
responders [10]. Moreover, higher levels of catecholamine analogue uptake before CRT
implantation are recorded in patients who will become responders to resynchronization
therapy, which would mean that the degree of ANS deterioration before CRT implantation
determines whether the patient will benefit from device implantation. In other words,
responders become those patients in whom the ANS is not structurally impaired but is in a
state of hibernation [10].
In heart failure patients, in addition to sinus rhythm, different types of arrhythmias
frequently occur, which limits the application of linear methods in the analysis of interbeat
interval time series [12]. Therefore, signals should be analyzed by nonlinear techniques,
and we applied the sample entropy approach [13] and as a complementary method of
detrended fluctuation analysis [14] to changes in cardiac and respiratory rhythm complexity
after CRT implantation. These techniques have been demonstrated to provide reliable
estimates of signal complexity in cardio-respiratory data [15–17].
We hypothesized that there were differences in cardio-respiratory parameters between
responders and non-responders to CRT. The aim of this study was to evaluate remote
changes in ANS function after CRT implantation, through examination of interactions
between cardiac and respiratory signals with complementary measures of time series
analysis, before and 9 months after device implantation. We aimed to determine whether
and in which direction the observed nonlinear parameters shift and whether their dynamics
differed between responders and non-responders to CRT. The second goal of our research is
to define the parameters of autonomic function that are capable to preoperatively separate
future responders from those who will not benefit from resynchronization therapy. Since
20–40% of patients do not respond to CRT [18], this could contribute to selection of patients
with heart failure who are more likely to benefit from this treatment.

2. Materials and Methods

2.1. Participants and Study Design
This is a one-group, quasi-experimental, pilot study, designed and conducted relying
on a template for intervention description and replication checklist and guide [19]. Baseline
measurements were performed in 54 patients. In the first year after device implantation,
three patients died, two from terminal heart failure and one from massive ischemic stroke.
Moreover, follow-up measurements were not performed in four patients because they
continued to regularly monitor the function of the device in the regional pacemaker
centers. Therefore, we evaluated 47 patients (14 females), mean (SD) age of 64.4 (8.7) years,
with heart failure and an indication for CRT device implantation according to current
guidelines [3]. Thus, all patients had left ventricular ejection fraction below 35%, left
bundle branch block (according to Strauss criteria [20]) on ECG, and all were on optimal
therapy for heart failure (Table 1). St. Jude Medical CRT devices were implanted in all
patients (St. Jude Medical Allure, St. Jude Medical Quadra Allure MP, St. Jude Medical
Quadra Assura MP). The study was approved by the Ethics Committee of the Faculty
of Medicine the University of Belgrade, and each subject signed an informed consent
(approval date: 17 March 2017, Ref. Numb.29/III-4). Measurements were performed before
(baseline) and approximately 9 months after CRT device implantation (follow-up). After
follow-up, patients were divided into two groups, responders (N = 27) and non-responders
(N = 20), in relation to the response to CRT, which was assessed according to changes in
certain clinical and echocardiographic parameters. More precisely, we defined that an
echocardiographic responder to CRT was a patient who increased LVEF by 5% or more
at follow-up, while clinical response was based on NYHA classification and a six-minute
walk test result and was considered positive if NYHA improvement was ≥1 class and
Entropy 2021, 23, 1126 3 of 15

distance covered during the test increased by 10%. In this study, the CRT responder had to
meet the stated echocardiographic and clinical criteria.

Table 1. Anthropometric data and baseline clinical parameters in CRT patients.

Parameter CRT Responders (%) CRT Non-Responders (%) p

Age (years) 62.8 ± 8.8 66.5 ± 8.3 0.20
Sex (male) 17 (62.9) 16 (80.0) 0.21
Atrial fibrillation 6 (22.2) 8 (40.0) 0.19
Ischemic heart disease 5 (18.5) 10 (50.0) 0.02
Arterial hypertension 21 (77.8) 17 (85.0) 0.53
Diabetes 7 (25.9) 4 (20.0) 0.64
Hyperlipidemia 14 (51.9) 12 (60.0) 0.58
Tobacco smoking 12 (44.4) 13 (65.0) 0.16
COPD 4 (14.8) 5 (25.0) 0.38
Chronic obstructive pulmonary disease—COPD.

2.2. Signals Measurements and Processing

Both experiments (baseline and follow-up) were conducted in the morning, between
7:00 and 8:00 a.m. in a quiet surrounding at the Pacemaker Center of the University
Clinical Center of Serbia. Baseline measurements were done immediately before device
implantation. For follow-up recording, CRT device was switched off, i.e., spontaneous
rhythm was used for cardiac signal examination. Data were acquired from 20 min of
ECG (RR intervals) and respiratory signal measurement of relaxed subjects in supine
position, breathing spontaneously, with a sampling frequency of 1000 Hz by the Biopac
MP100 system and AcqKnowledge 3.9.1 software (BIOPAC System, Inc., Santa Barbara,
CA, USA). The ECG data were collected using the ECG 100C electrocardiogram amplifier
module. The RSP 100C respiratory pneumogram amplifier module with TSD 201 transducer
attached to the belt (adjustable nylon strap) was used to measure abdominal expansion and
contraction [21]. The interbeat interval time series in patients with sinus rhythm and sinus
rhythm with ventricular extrasystoles and with atrial fibrillation were analyzed. Interbeat
(RR) intervals and interbreath (BB) intervals were extracted from recorded signals using
tool Pick Peaks from OriginPro 8.6 (OriginLab Corporation, Northampton, MA, USA).
Heart rate (HR) and breathing frequency (BF) were obtained as a reciprocal value of subject
mean RR interval and mean BB interval. The resampling procedures were performed
to obtain equal equidistant resampled time series (time series with the same number of
equidistant samples) of RR intervals and respiratory signals [15].

2.3. Sample Entropy and Cross Sample Entropy

Sample entropy (SampEn) and cross-sample entropy (crossSampEn) are non-linear
measures of time series regularity derived from the probability of finding a similar pattern
within signal/signals [13]. It is defined as the negative natural logarithm of the conditional
probability that two sequences similar for m points remain similar within tolerance r at
the next point, where self-matches are not included. We calculated SampEn from RR
interval series (SampEnRR) and corresponding respiratory signal (SampEnResp). The
signal with a small number of similar patterns is characterized by higher values of SampEn
which indicates its greater unpredictability (irregularity). Similarly, coupling of two signals
with a small number of similar patterns would result in high values of crossSampEn
indicating their high asynchrony i.e., low association between analyzed systems. SampEn
and crossSampEn were calculated with fixed input variables m = 2 (window size) and
r = 0.15·SD (tolerance, SD—standard deviation of time series) from whole time series
(approximately about 1200 samples). Previously developed MATLAB scripts were used
to calculate SampEn and crossSampEn from normalized and two equally equidistant
resampled series of RR intervals and of the respiratory signal [15]. Entropy measures and
scaling exponents were computed using MATLAB (R2007b, TheMathWorks, Inc., Natick,
MA, USA).
Entropy 2021, 23, 1126 4 of 15

2.4. Detrended Fluctuation Analysis (DFA)

Detrended fluctuation analysis was used to calculate short-term α1 and long-term α2 of
the equally equidistant resampled series of RR intervals and respiratory signal data [14,20].
In general, for the time series x(i), of length N, one calculates the profile y(k):

y(k) = ∑ ik=1 [xi − x] (1)

The profile y(k) is divided into non-overlapping segments of length n. In each segment,
the local trend is the linear least squares fit for the segment data. The squared fluctuation
of the segment is calculated from detrended walk which is the difference between y(k)
and the local trend yn (k). The root-mean-square fluctuation function of the time series is
determined by averaging over all segments:
r i2
1 h
F (n) =
N ∑ kN=1 y ( k ) − yn(k) (2)

The scaling exponent is defined as the slope of the linear regression line of log(F(n))
against log(n). We found one crossover, i.e., two different scaling exponents for both time
series at n = 16. The value of scaling exponents indicates correlation properties: α < 0.5
(anticorrelated), α ~ 0.5 (uncorrelated—white noise), α ~ 1 (correlated—1/f noise), and
α ~ 1.5 (strongly correlated—Brownian noise). Detrended fluctuation analysis typically
shows two ranges of scale invariance, which are quantified with two separate scaling
exponents, α1 and α2 , reflecting the short-term and long-term correlations, respectively.
Previously developed MATLAB scripts were used to calculate scaling exponents from
normalized and two equally equidistant resampled series of RR intervals and of the
respiratory signal [22].

2.5. Statistical Analysis

At the beginning of the study, we calculated the sample size with a population of
54 patients, a confidence level of 95% and 5% of confidence interval. Using these parameters,
we found that the estimated recommended sample size was 47 patients.
Mixed design ANOVA for repeated measures was applied to find significant group x
measures interaction. Separately, because we analyzed two groups of patients (responders
and non-responders to CRT) with repeated measurements, we used one-way ANOVA
for intergroup comparisons at the baseline and at the follow-up measures. On the other
side, we used repeated measures ANOVA for intragroup comparisons between baseline
and follow-up data. The results of intergroup comparisons are presented in Table 2,
and statistically significant results of intra- and intergroup comparisons are presented in
Figures.
We estimated effect sizes, as a measure of magnitude of the difference between re-
sponders and non-responders groups by Cohen’s term ds for all calculated parameters and
presented the results in Table 2. In computing, we used pooled standard deviation for both
groups of parameters obtained in follow-up measurement. Cohen classified effect sizes as
small (d = 0.2), medium (d = 0.5) and large (d ≥ 0.8) [23].
We applied ROC analysis to examine the diagnostic ability of the ratio of CrossSampEn
value after and before CRT to discriminate patients with successful resynchronization
therapy. We calculated the area under the ROC curve that can be used as a criterion for
measuring the discriminative ability of variables in a given clinical situation.
Statistical analyses were performed using the software package SPSS Statistics (version
17.0, SPSS Inc., Chicago, IL, USA). For all analyses, probability values less than or equal to
0.05 were considered as statistically significant.
Entropy 2021, 23, 1126 5 of 15

Table 2. Clinical parameters and parameters of analysis of cardiac, respiratory rhythm and cardio-respiratory interactions
with comparison between responders and non-responders to cardiac resynchronization therapy (CRT) from measurements
before (baseline) and after (follow-up) device implantation.

Baseline Follow-Up
Responders Non-Responders Responders Non-Responders
p p Cohen’s ds
(N = 27) (N = 20) (N = 27) (N = 20)
BiVP (%) 96 ± 0.7 95 ± 1 0.85 0.25
LVEF (%) 25 ± 2 26 ± 2 0.38 43 ± 2 24 ± 1 0.04 2.01
NYHA 2.4 ± 0.1 2.6 ± 0.1 0.19 1.6 ± 0.1 2.6 ± 0.1 0.01 1.67
6MWT 280 ± 10 230 ± 20 0.34 360 ± 20 230 ± 20 0.01 1.92
HR (beat/min) 73 ± 2 70 ± 3 0.44 64 ± 2 64 ± 3 0.84 0.06
BF (Hz) 0.27 ± 0.01 0.29 ± 0.01 0.17 0.25 ± 0.02 0.30 ± 0.01 0.01 0.14
SampEnRR 1.31 ± 0.08 1.6 ± 0.2 0.10 1.34 ± 0.08 1.5 ± 0.2 0.30 0.31
SampEnResp 1.49 ± 0.07 1.30 ± 0.09 0.09 1.50 ± 0.06 1.56 ± 0.07 0.55 0.18
CrossSampEn 2.4 ± 0.2 2.2 ± 0.1 0.35 2.3 ± 0.1 3.3 ± 0.4 0.01 0.76
α1 (RR) 0.80 ± 0.07 0.72 ± 0.07 0.43 0.83 ± 0.07 0.60 ± 0.07 0.01 0.77
α2 (RR) 0.84 ± 0.03 0.75 ± 0.04 0.04 0.79 ± 0.04 0.74 ± 0.06 0.43 0.23
α1 (Resp) 0.35 ± 0.04 0.35 ± 0.06 0.97 0.43± 0.04 0.27 ± 0.04 0.01 0.83
α2 (Resp) 0.35 ± 0.05 0.44 ± 0.06 0.26 0.59 ± 0.03 0.59 ± 0.06 0.96 0.16
Biventricular pacing—BiVP, Left ventricular ejection fraction—LVEF, six-minute walk test—6MWT, heart rate—HR, breathing frequency—
BF, sample entropy of RR interval series—SampEnRR, sample entropy of respiratory signal time series—SampEnResp, cross-sample
entropy—CrossSampEn, short-term scaling exponent of RR interval series—α1 (RR), long-term scaling exponent of RR interval series—
α2 (RR), short-term scaling exponent of respiratory signal series—α1 (Resp), long-term scaling exponent of respiratory signal series—α2 (Resp).
p—significance test result for intergroup comparisons (responders vs. non-responders). Cohen’s ds —measure of effect size. Values are
mean ± standard error.

3. Results
Table 1 shows demographic and clinically relevant, baseline data of CRT patients.
Baseline clinical parameters did not differ significantly between groups of future CRT
responders and non-responders, except that there were significantly more patients with
ischemic heart disease among patients who would not benefit from resynchronization
therapy (Table 1).
Clinical and echocardiographic examination before follow-up signal recording showed
a statistically significant difference in left ventricular ejection fraction, NYHA class, and
distance covered on a six-minute walk test between CRT responders and non-responders
(Table 2).
We found that in all patients, resting heart rate decreased significantly after CRT
implantation (Figure 1), but the complexity of the RR intervals, measured by SampEnRR,
did not change. However, in CRT non-responders SampEnRR was higher in both the
baseline and follow-up measurements, but the differences were not statistically significant
(Table 2, Figure 1).
Entropy 2021, 23, 1126 6 of 15

Figure 1. Cardiac and respiratory signal analysis parameters in CRT responders and non-responders, before (baseline) and
after (follow-up) device implantation. Heart rate—HR (A), breathing frequency—BF (B), sample entropy of RR interval
series—SampEnRR (C), and sample entropy of respiratory signal time series—SampEnResp (D). ** p < 0.01, * p < 0.05
intragroup comparison (F-up vs. Baseline), ## p < 0.01 intergroup comparison (Responders F-up vs. Non-responders F-up).
Values are mean + standard error.

By additional complementary measures, short-term, α1 (RR), and long-term, α2 (RR),

scaling exponents of RR intervals, some of these differences were statistically determined
(Figure 2). Namely, both scaling exponents were higher in CRT responders, which indicated
a more correlated RR interval time series or more regular time series than in non-responders.
Statistically significant differences between groups were achieved for α1 (RR) at follow up
measurements and for α2 (RR) at baseline measurements (Figure 2). The long-term scaling
exponent of RR interval series was the only parameter that could preoperatively separate
future responders from those who would not benefit from resynchronization therapy
(Figure 2).
Entropy 2021, 23, 1126 7 of 15

Figure 2. Cardiac and respiratory signal parameters obtained by detrended fluctuation analysis in CRT responders and
non-responders, before (baseline) and after (follow-up) device implantation. Short-term scaling exponent of RR interval
series—α1 (RR) (A), short-term scaling exponent of respiratory signal series—α1 (Resp) (B), long-term scaling exponent of RR
interval series—α2 (RR) (C), and long-term scaling exponent of respiratory signal series—α2 (Resp) (D). ** p < 0.01, * p < 0.05
intragroup comparisons (F-up vs. Baseline), ## p < 0.01 Responders F-up vs. Non-responders F-up, # p < 0.05 Responders
Baseline vs. Non-responders Baseline, intergroup comparisons. Values are mean + standard error.

There was no significant difference between CRT responders and non-responders in

breathing frequency at the baseline measurement (Figure 1). Compared to the baseline
recording, at the follow-up examination in responders to CRT, respiratory rate was sig-
nificantly lower, while in non-responders, it was slightly higher. These findings resulted
in a statistically significant difference in breathing frequency between responders and
non-responders after follow-up measurements (Figure 1). In CRT non-responders, respi-
ratory rhythm complexity quantified by sample entropy, SampEnResp, at baseline was
significantly lower than after CRT implantation and slightly lower than in responders at
both measurements (Figure 1). Respiratory short-term, α1 (Resp), and long-term, α2 (Resp),
scaling exponents had the same direction of changes in all patients with a statistically
significant increase in CRT responders and only for α2 (Resp) in non-responders (Figure 2).
Exceptional behavior at follow-up measurements was found for α1 (Resp) in CRT non-
responders, where there was no increase in the correlation of respiratory time series, as
in other cases after CRT implantation. These findings probably reflected the absence of
some respiratory control mechanisms over short scales in CRT non-responders that may
differentiate responders from non-responders.
In Figure 3, we presented the changes of CrossSampEn between RR interval and
respiratory signal time series in this study. In CRT responders, CrossSampEn did not
change significantly between baseline and follow-up measurements. Contrary, in non-
responders, after CRT implantation, CrossSampEn was statistically significantly reduced
Entropy 2021, 23, 1126 8 of 15

compared to the baseline examination, achieving values that were statistically significantly
higher than in CRT responders (Figure 3).

Figure 3. Cross sample entropy between RR interval series and respiratory signal series (CrossSampEn) in CRT responders
and non-responders, before (baseline) and after (follow-up) device implantation. * p < 0.05 F-up vs. Baseline, ## p < 0.01
Responders F-up vs. Non-responders F-up. Values are mean + standard error.

Further, we calculated the ROC curve values for the ratio (rCrossSampEn) defined as
the quotient of CrossSampEn obtained for follow-up and baseline measurements (Figure 4).
According to the characterization of ROC curves, our result of the area under the ROC
curve belongs to the fair category and rCrossSampEn was not as strong a discriminator as
we expected, but this result may indicate the direction for application of similar parameters
in further studies.
It is important to emphasize that the underlying cardiac rhythm did not change in any
of observed patients at the follow-up compared to the baseline examination.
Entropy 2021, 23, 1126 9 of 15

Figure 4. ROC curve for the ratio of cross sample entropy obtained after and before CRT. rCrossSampEn =
(CrossSampEn)F-up /(CrossSampEn)Baseline .

4. Discussion
To our knowledge, this is the first study designed and conducted to determine the
effect of cardiac resynchronization therapy on the relationship between the cardiovascular
and respiratory systems in patients with heart failure. In this research, we used available
methods that have been shown to provide the best assessment of cardiac and respiratory
signal complexity.
In CRT responders, 9 months after device implantation, the resting heart rate is signifi-
cantly reduced. This is the result of restoring autonomic balance, primarily the recovery
of vagal activity that is predominantly responsible for resting heart rate control [24]. This
finding confirms that the clinical progress in patients with heart failure, i.e., NYHA class im-
provement, is associated with a lower resting rate [24]. Moreover, after CRT implantation,
antiarrhythmics are more freely prescribed because there is no longer a fear of bradycardia
developing, and in patients with atrial fibrillation as the underlying rhythm, higher doses
of these drugs are necessary to achieve the target percentage of biventricular pacing. More
intensive antiarrhythmic therapy is actually the main reason why CRT non-responders
also had lower resting ventricular rate at follow-up compared to baseline examination.
Moreover, all patients were one year older at the follow-up, and we know that with each
year of life the heart rate decreases [25]. An indicator of better autonomic control and
baroreflex activity in patients who had a favorable response to resynchronization therapy
is that they have a higher complexity of RR intervals, i.e., a more irregular heart rhythm
on the follow-up. Indeed, baseline and follow-up values of SampEnRR do not differ sta-
tistically significantly, but it is clear that the values of this parameter in CRT responders
and non-responders move in different directions on the control recording. Certainly, the
optimization of therapy for heart failure after CRT implantation also plays an important
Entropy 2021, 23, 1126 10 of 15

role, as there is evidence that ß-blockers, some ACE inhibitors, and aldosterone antagonists
lead to an increase in vagal tone [26–28]. Therefore, the improvement in cardiac neuroad-
renergic, parasympathetic, and baroreflex function due to cardiac resynchronization and
optimization of drug therapy leads to an increase in heart rate variability and heart rate
turbulence, which results in more irregular heart rhythm with lower resting rate.
In CRT responders, a significantly lower respiratory rate was registered at the control
examination compared to the baseline recording, which resulted in a statistically significant
difference in the values of this parameter between responders and non-responders after
follow-up measurements. Respiratory rate directly depends on the extent of stimulation
of peripheral and central chemoreceptors. In CRT responders, cardiac systolic function
recovers; oxygen delivery to the periphery increases, i.e., better blood flow is achieved in
organs, especially in the kidneys, which reduces hypoxia and metabolic acidosis, result-
ing in reduced chemoreceptor stimulation and reduced respiratory rate [29–31]. In heart
failure, there is an increased sensitivity of peripheral chemoreceptors to a drop in partial
pressure of oxygen, and it is assumed that their sensitivity decreases with the improvement
of cardiac function [32]. In CRT responders, it is also likely (at least in part) that there
will be a reduction in respiratory muscle weakness. This would mean a decrease in the
production of specific metabolites, which leads to a decrease in the activity of muscle
parenchyma chemoreceptors, i.e., a muscle metaboreflex activity reduction, which con-
tributes to a decrease in sympathetic tone, respiratory rate, dyspnea and improved aerobic
function [33,34]. In future studies, it would be interesting to examine the joint contribution
of resynchronization therapy and inspiratory muscle training, which we already know
reduces the respiratory muscle metaboreflex. A significant reduction in respiratory rate
during follow-up in CRT responders is a confirmation of the previously shown relationship
of respiratory rate with NYHA class and left ventricular ejection fraction [35]. Thus, in
patients who have benefited from CRT implantation, i.e., in whom there was a decrease
in the NYHA class and an increase in the left ventricular ejection fraction, a reduction
in breathing frequency was registered. In previous research, we have shown that in the
presence of impaired autonomic control, a more regular heart rhythm is related to a higher
respiratory rate [35]. In the present study, in patients who are non-responders to resynchro-
nization therapy, with a pronounced imbalanced autonomic nervous system, we found
a reduction in complexity of heart rhythm and an increase in breathing frequency at the
follow-up examination.
It is interesting that in CRT responders the respiratory rhythm was more irregular
at the follow-up compared to baseline recording, with no statistical significance, and
in CRT, non-responders statistically significantly more irregular. In addition, patients
who would have a favorable response to resynchronization therapy had higher values of
SampEnResp before device implantation, and this was on the edge of statistical significance.
It is difficult to analyze the regularity of respiratory rhythm in patients with heart failure,
especially based on a relatively short record during wakefulness, because it is influenced
by many factors, such as age, sex, left ventricular ejection fraction, NYHA class, and
presence of breathing disorders [36,37]. It has been shown that the frequency of respiratory
disorders is lower in women and patients on beta-blocker therapy, while the prevalence
of these disorders during sleep and wakefulness increases with deteriorating in cardiac
function [36,37]. In previous research, we have concluded that among patients with heart
failure, those who have atrial fibrillation as the underlying rhythm have the highest values
of SampEnResp [17]. In the same study, we have concluded that respiratory rhythm is
more regular in patients with heart failure and without atrial fibrillation than in healthy
controls [17]. Thus, the fact that, prior to device implantation, future CRT responders
compared to non-responders have significantly higher values of SampEnResp probably
indicates that they have a lower degree of structural impairment of ANS, as well as that
they still have preserved compensatory mechanisms, which have not yet been sufficiently
examined and understood. The increase in respiratory rhythm irregularity at follow-up
in patients who did not respond favorably to CRT can be explained by the more frequent
Entropy 2021, 23, 1126 11 of 15

occurrence of respiratory disorders in these patients, due to further deterioration of cardiac

function but also by the fact that male and patients with atrial fibrillation were more often
CRT non-responders in our study.
Synchronization of cardiac and respiratory rhythms slightly increased after device
implantation in CRT responders, while in non-responders it was statistically significantly
reduced compared to the baseline examination, achieving values of CrossSampEn that
were statistically significantly higher in these patients than in CRT responders. Data on car-
diopulmonary synchronization in heart failure are deficient in the literature. Furthermore,
unlike respiratory sinus arrhythmia, for this type of cardio-respiratory interaction, we
do not know enough about underlying physiologic mechanism nor about its significance
even in healthy subjects [38,39]. Cardiopulmonary synchronization is considered to be
of great importance for fine matching of ventilation and perfusion and thus maximizing
energy utilization [39]. It has been shown that synchronization of cardiac and respiratory
rhythms in swimmers and athletes is superior to that in ordinary people, i.e., that car-
diopulmonary enhancement represents a better physiological state [38,40]. Cardiac and
respiratory rhythms are directly controlled by the ANS, but also with influences from the
central nervous system [39]. In our previous study, we concluded that cardiopulmonary
synchronization, despite the existence of ANS imbalance, is maintained in patients with
heart failure who are in sinus rhythm, probably due to powerful effect of compensatory
mechanisms [17]. However, the results of this study indicated that with the progression of
heart failure, with a further increase in sympathetic activity, the compensatory mechanisms
lose their importance. In other words, with further deterioration of autonomic control, the
respiratory system is no longer capable to adapt to changes in heart rhythm, leading to a
loss of synchrony in the cardio and respiratory coupled systems.
The autonomic nervous system, more precisely baroreflex activity, is responsible for
the short-term regulation of heart rate and blood pressure [41]. Findings from previous
studies have indicated a strong negative correlation of the short-term scaling exponent
of RR interval series with the level of sympathetic activity, i.e., with the plasma levels of
circulating catecholamines [42]. Therefore, it is expected that in heart failure, accompanied
by strong sympathetic activation, a decrease in the value of this exponent is recorded [43].
Moreover, it has been shown that lower α1 (RR) is associated with increased mortality and
hospitalization rate in patients with heart failure and that this exponent is independent
predictor of heart failure [43,44]. We found that in CRT responders, nine months after
device implantation, α1 (RR) increases slightly but reaches values that are statistically
significantly higher than in non-responders. Therefore, this difference in α1 (RR) between
responders and non-responders is primarily due to a further decrease in α1 (RR) in patients
in whom the deterioration of cardiac function was not stopped, rather than an increase in
the value of this parameter in CRT responders. CRT restores autonomic balance leading
to ß-adrenergic signaling depression. However, recent findings from animal models and
few clinical studies have suggested that restoring sympathovagal balance also involves
modulation of cholinergic activity [45]. This explains the statistically insignificant increase
in α1 (RR) in CRT responders in our study. Restoration of ANS balance in these patients
implies a reduction of sympathetic activity, which leads to an increase in α1 (RR) but
also an improvement in vagal tone, which is associated with a decrease in the value
of this parameter [46]. In CRT non-responders, a further increase in sympathetic and
withdrawal in vagal activity, with an increase in respiratory rate which additionally reduces
α1 (RR), leads to a significant decrease in the value of this exponent at the follow-up
examination [47,48]. Knowing that α1 (RR) is strongly associated with the breathing pattern,
obtained results suggest that CRT non-responders developed alterations in cardiac rhythm
related to respiration during follow-up.
The long-term scaling exponent of RR interval series has been less studied in heart fail-
ure, but the results of several studies have shown that its values are statistically significantly
higher in patients with heart failure with reduced ejection fraction compared to healthy
controls [43,49,50]. In this work, we found a slight reduction of α2 (RR) at follow-up exami-
Entropy 2021, 23, 1126 12 of 15

nation in CRT responders. It is not a surprising result, because we know that a decrease
in sympathetic tone leads to a decrease in α2 (RR) and that the values of this parameter in
patients with heart failure with preserved ejection fraction (many CRT responders become
part of that subpopulation) do not differ significantly compared to healthy subjects [41,46].
However, we expected statistically significant changes in the value of this exponent both in
responders and non-responders, especially since this is the only parameter that separated
these two groups of patients at the baseline examination. Our results indicate that α2 RR
should not be considered only as an indicator of the state of ANS, i.e., the degree of block-
ade of the ß receptors but also as a marker of preservation of the function of the sinoatrial
node cells [51]. In our study, in CRT non-responders, α2 (RR) practically does not change, so
the values of this parameter between responders and non-responders at the follow-up do
not differ significantly due to the decrease in α2 (RR) in CRT responders. Therefore, lower
α2 (RR) in patients with heart failure should be definitely understood both as an index of
reduced sympathetic activity and as a sign of sinus node dysfunction. This would also
mean that the success of resynchronization therapy in conditions of impaired function
of sinoatrial node cells is limited. Of course, the baseline values of α2 (RR), as well as the
dynamics of this parameter during follow-up, should also be analyzed from the aspect of
preserving some slower physiological regulatory mechanisms, probably neuroautonomic,
which have not yet been fully examined [50].
Fractal organization of breathing patterns in adults is much less examined, and in
pathological conditions, it is limited almost exclusively to respiratory diseases [52,53].
Certainly, the respiratory time series also have nonstationary characteristics, and nonlinear
methods, such as detrended fluctuation analysis, are needed to quantify their scaling
behavior. So far, we know that long-range correlations are present in human respiratory
dynamics and that can be reduced in elderly men [52]. It is considered that in physiologi-
cal conditions, there is a statistically significant difference between values of short- and
long-term scaling exponent of respiratory signal series, with dominance of α1 (Resp) and
values of α2 (Resp) in the range of anticorrelations, which indicates the coordinated activity
of multiple regulatory mechanisms of the respiratory rhythm [22]. Raoufy et al. have
shown that in asthmatic patients compared to healthy controls, long-range correlations
are significantly decreased and that nonlinear analysis of respiratory rate variability in
these patients has a diagnostic value and can be useful in differentiating uncontrolled from
controlled and non-atopic from atopic asthma [53]. In our study, at baseline examination,
future CRT responders had almost identical values of α1 (Resp) and α2 (Resp), and in non-
responders, long-term scaling exponent had slightly higher values. During follow-up, there
was a statistically significant increase in the values of both α1 (Resp) and α2 (Resp) in CRT
responders, while in non-responders with a significant increase in α2 (Resp), a significant
decrease in α1 (Resp) was registered 9 months after device implantation. An explanation of
the dynamics of these scaling exponents in patients who responded unfavorably to resyn-
chronization therapy should probably be sought in the augmented sympathetic activity,
reduced vagal tone and increased respiratory rate in these patients. In CRT responders,
despite the increase in α1 (Resp), the values of this exponent remain lower than α2 (Resp) at
the follow-up examination. This result is an indication that we still do not know enough
about the interactions over multiple time scales of different control and feedback systems
but also that the fact that the patient has responded favorably to resynchronization therapy
does not mean that he no longer has heart failure, i.e., that the ANS imbalance is no longer
present [52].
During follow-up, from control to control, there were changes in cardiovascular medi-
cations, in relation to the achieved improvement or deterioration of cardiac function, the
percentage of biventricular pacing, and the occurrence of atrial and ventricular arrhythmias.
However, these changes were mainly related to the doses of administered drugs, rather
than to the introduction of new groups of drugs. The effect of these changes, primarily
from beta blockers (which have been shown to not only reduce sympathetic activity but
Entropy 2021, 23, 1126 13 of 15

also increase vagal tone), on the examined parameters of cardiac and respiratory signal
complexity was not analyzed, and this is a limitation of our study.

5. Conclusions
In this study, the complexity and synchrony of cardiac and respiratory signals after
cardiac resynchronization therapy were examined, and a significant association between
CRT success and cardio-respiratory coupling stability was demonstrated. We showed that
the entropy values of the RR intervals were higher, although statistically insignificantly, in
CRT non-responders both before and after device implantation. Both scaling exponents
of RR interval series were higher in CRT responders at the baseline examination, and
α2 (RR) was the only parameter that could preoperatively separate future responders from
those who would not benefit from resynchronization therapy. At the follow-up recording,
in responders to CRT, α1 (RR) increased slightly but reached values that statistically are
significantly higher than in non-responders. Respiratory scaling exponents increased at
follow-up in both groups, with the exception of α1 (Resp) whose values were significantly
reduced in CRT non-responders. Synchronization of cardiac and respiratory rhythms was
preserved after device implantation only in CRT responders.

Author Contributions: Conceptualization, M.M.P., S.U.P. and N.N.R.; methodology, M.M.P.; valida-
tion, S.U.P.; formal analysis, M.M.P.; investigation, N.N.R.; data curation, N.N.R.; writing—original
draft preparation, N.N.R. and M.M.P.; writing—review and editing, N.N.R., M.M.P., S.U.P. and G.M.;
visualization, M.M.P.; supervision, G.M. and S.U.P. All authors have read and agreed to the published
version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: The study was conducted according to the guidelines of
the Declaration of Helsinki, and approved by the Institutional Ethics Committee of the Faculty of
Medicine the University of Belgrade (date of approval: 17 March 2017, Ref. Numb.29/III-4).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the
study.
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. O’Brien, T.; Park, M.S.; Youn, J.C.; Chung, E.S. The past, present and future of cardiac resynchronization therapy. Korean Circ. J.
2019, 49, 384–399. [CrossRef]
2. Savarese, G.; Lund, L.H. Global Public Health Burden of Heart Failure. Card. Fail. Rev. 2017, 3, 7–11. [CrossRef]
3. Ponikowski, P.; Voors, A.A.; Anker, S.D.; Bueno, H.; Cleland, J.G.; Coats, A.J.; Falk, V.; González-Juanatey, J.R.; Harjola, V.P.;
Jankowska, E.A.; et al. 2016 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: The Task Force for
the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC). Developed with the
special contribution of the Heart Failure Association (HFA) of the ESC. Eur. J. Heart. Fail. 2016, 18, 891–975. [CrossRef] [PubMed]
4. Daubert, C.; Behar, N.; Martins, R.P.; Mabo, P.; Leclercq, C. Avoiding non-responders to cardiac resynchronization therapy: A
practical guide. Eur. Heart J. 2017, 38, 1463–1472. [CrossRef] [PubMed]
5. Jafferani, A.; Leal, M.A. Advances in cardiac resynchronization therapy. J. Innov. Card. Rhythm Manag. 2019, 10, 3681–3693.
[CrossRef] [PubMed]
6. Boriani, G.; Regoli, F.; Saporito, D.; Martignani, C.; Toselli, T.; Biffi, M.; Francolini, G.; Diemberger, I.; Bacchi, L.; Rapezzi, C.; et al.
Neurohormones and inflammatory mediators in patients with heart failure undergoing cardiac resynchronization therapy: Time
courses and prediction of response. Petides 2006, 27, 1776–1786. [CrossRef] [PubMed]
7. Dong, Y.X.; Burnett, J.C., Jr.; Chen, H.H.; Sandberg, S.; Yang, Y.Z.; Zhang, Y.; Chen, P.S.; Cha, Y.M. Effect of cardiac resynchroniza-
tion therapy on broad neurohormone biomarkers in heart failure. J. Interv. Card. Electrophysiol. 2011, 30, 241–249. [CrossRef]
[PubMed]
8. Asgardoon, M.H.; Vasheghani-Farahani, A.; Sherafati, A. Usefulness of biomarkers for predicting response to cardiac resynchro-
nization therapy. Curr. Cardiol. Rev. 2020, 16, 132–140. [CrossRef] [PubMed]
9. Florea, V.G.; Cohn, J.N. The autonomic nervous system and heart failure. Circ. Res. 2014, 114, 1815–1826. [CrossRef]
Entropy 2021, 23, 1126 14 of 15

10. Martignani, C.; Diemberger, I.; Nanni, C.; Biffi, M.; Ziacchi, M.; Boschi, S.; Corzani, A.; Fanti, S.; Sambuceti, G.; Boriani, G. Cardiac
resynchronization therapy and cardiac sympathetic function. Eur. J. Clin. Investig. 2015, 45, 792–799. [CrossRef]
11. Scholtens, A.M.; Braat, A.J.; Tuinenburg, A.; Meine, M.; Verberne, H.J. Cardiac sympathetic innervation and cardiac resynchro-
nization therapy. Heart Fail. Rev. 2014, 19, 567–573. [CrossRef]
12. Platiša, M.M.; Radovanović, N.N.; Milašinović, G.; Pavlović, S.U. Sample Entropy Approach to the Examination of Cardio-
Respiratory Coupling in Response to Cardiac Resynchronization Therapy. In Proceedings of the Conference Entropy 2021: The
Scientific Tool of the 21st Century, Online, 5–7 May 2021; MDPI: Basel, Switzerland, 2021. [CrossRef]
13. Richman, J.S.; Moorman, J.R. Physiological time-series analysis using approximate entropy and sample entropy. Am. J. Physiol.
Heart Circ. Physiol. 2000, 278, H2039–H2049. [CrossRef]
14. Peng, C.K.; Havlin, S.; Stanley, H.E.; Goldberger, A.L. Quantification of scaling exponents and crossover phenomena in nonsta-
tionary heartbeat time series. Chaos 1995, 5, 82–87. [CrossRef]
15. Kapidžić, A.; Platiša, M.M.; Bojić, T.; Kalauzi, A. Nonlinear properties of cardiac rhythm and respiratory signal under paced
breathing in young and middle-aged healthy subjects. Med. Eng. Phys. 2014, 36, 1577–1584. [CrossRef]
16. Platiša, M.M.; Radovanović, N.N.; Kalauzi, A.; Milašinović, G.; Pavlović, S.U. Multiscale entropy analysis: Application to
cardio-respiratory coupling. Entropy 2020, 22, 1042. [CrossRef]
17. Radovanović, N.N.; Pavlović, S.U.; Milašinović, G.; Kirćanski, B.; Platiša, M.M. Bidirectional cardio-respiratory interactions in
heart failure. Front. Physiol. 2018, 9, 165. [CrossRef] [PubMed]
18. Dhesi, S.; Lockwood, E.; Sandhu, R.K. Troubleshooting cardiac resynchronization therapy in nonresponders. Can. J. Cardiol. 2017,
33, 1060–1065. [CrossRef]
19. Hoffmann, T.C.; Glasziou, P.P.; Boutron, I.; Milne, R.; Perera, R.; Moher, D.; Altman, D.G.; Barbour, V.; Macdonald, H.; Johnston,
M.; et al. Better reporting of interventions: Template for intervention description and replication (TIDieR) checklist and guide.
BMJ. 2014, 348, g1687. [CrossRef] [PubMed]
20. Strauss, D.G.; Selvester, R.H.; Wagner, G.S. Defining left bundle branch block in the era of cardiac resynchronization therapy. Am.
J. Cardiol. 2011, 107, 927–934. [CrossRef]
21. Platiša, M.M.; Bojić, T.; Pavlović, S.U.; Radovanović, N.N.; Kalauzi, A. Uncoupling of cardiac and respiratory rhythm in atrial
fibrillation. Biomed. Eng. Biomed. Tech. 2016, 61, 657–663. [CrossRef] [PubMed]
22. Zeković, J.; Madžgalj, Š.; Platiša, M.M. Detrended fluctuation analysis of heart and respiratory rhythm in atrial fibrillation. In
Proceedings of the Computing in Cardiology Conference, Maastricht, The Netherlands, 23–26 September 2018. [CrossRef]
23. Sullivan, G.M.; Feinn, R. Using Effect Size-or Why the P Value Is Not Enough. J. Grad. Med. Educ. 2012, 4, 279–282. [CrossRef]
24. Hori, M.; Okamoto, H. Heart rate as a target of treatment of chronic heart failure. J. Cardiol. 2012, 60, 86–90. [CrossRef]
25. Bonnemeier, H.; Richardt, G.; Potratz, J.; Wiegand, U.K.; Brandes, A.; Kluge, N.; Katus, H.A. Circadian profile of cardiac
autonomic nervous modulation in healthy subjects: Differing effects of aging and gender on heart rate variability. J. Cardiovasc.
Electrophysiol. 2003, 14, 791–799. [CrossRef]
26. Aronson, D.; Burger, A.J. Effect of beta-blockade on heart rate variability in decompensated heart failure. Int. J. Cardiol. 2001, 79,
31–39. [CrossRef]
27. Kontopoulos, A.G.; Athyros, V.G.; Papageorgiou, A.A.; Skeberis, V.M.; Basayiannis, E.C.; Boudoulas, H. Effect of angiotensin-
converting enzyme inhibitors on the power spectrum of heart rate variability in post-myocardial infarction patients. Coron. Artery
Dis. 1997, 8, 517–524. [PubMed]
28. Yee, K.M.; Pringle, S.D.; Struthers, A.D. Circadian variation in the effects of aldosterone blockade on heart rate variability and QT
dispersion in congestive heart failure. J. Am. Coll. Cardiol. 2001, 37, 1800–1807. [CrossRef]
29. Urso, C.; Brucculeri, S.; Caimi, G. Acid-base and electrolyte abnormalities in heart failure: Pathophysiology and implications.
Heart Fail. Rev. 2015, 20, 493–503. [CrossRef]
30. Di Lullo, L.; Bellasi, A.; Barbera, V.; Russo, D.; Russo, L.; Di Iorio, B.; Cozzolino, M.; Ronco, C. Pathophysiology of the cardio-renal
syndromes types 1-5: An uptodate. Indian Heart J. 2017, 69, 255–265. [CrossRef] [PubMed]
31. Konishi, M.; Akiyama, E.; Suzuki, H.; Iwahashi, N.; Maejima, N.; Tsukahara, K.; Hibi, K.; Kosuge, M.; Ebina, T.; Sakamaki, K.;
et al. Hypercapnia in patients with acute heart failure. ESC Heart Fail. 2015, 2, 12–19. [CrossRef] [PubMed]
32. Marcus, N.J.; Del Rio, R.; Schultz, H.D. Central role of carotid body chemoreceptors in disordered breathing and cardiorenal
dysfunction in chronic heart failure. Front. Physiol. 2014, 5, 438. [CrossRef]
33. Fernandez-Rubio, H.; Becerro-de-Bengoa-Vallejo, R.; Rodríguez-Sanz, D.; Calvo-Lobo, C.; Vicente-Campos, D.; Chicharro, J.L.
Inspiratory Muscle Training in Patients with Heart Failure. J. Clin. Med. 2020, 9, 1710. [CrossRef]
34. Fernández-Rubio, H.; Becerro-de-Bengoa-Vallejo, R.; Rodríguez-Sanz, D.; Calvo-Lobo, C.; Vicente-Campos, D.; Chicharro, J.L.
Unraveling the Role of Respiratory Muscle Metaboloreceptors under Inspiratory Training in Patients with Heart Failure. Int. J.
Environ. Res. Public Health 2021, 18, 1697. [CrossRef] [PubMed]
35. Forleo, G.B.; Santini, L.; Campoli, M.; Malavasi, M.; Scaccia, A.; Menichelli, M.; Riva, U.; Lamberti, F.; Carreras, G.; Orazi, S.;
et al. Long-term monitoring of respiratory rate in patients with heart failure: The Multiparametric Heart Failure Evaluation
in Implantable Cardioverter-Defibrillator Patients (MULTITUDE-HF) study. J. Interv. Card. Electrophysiol. 2015, 43, 135–144.
[CrossRef]
36. Pietrock, C.; von Haehling, S. Sleep-disordered breathing in heart failure: Facts and numbers. ESC Heart Fail. 2017, 4, 198–202.
[CrossRef]
Entropy 2021, 23, 1126 15 of 15

37. Silva, R.S.; Figueiredo, A.C.; Mady, C.; Lorenzi-Filho, G. Breathing disorders in congestive heart failure: Gender, etiology and
mortality. Braz. J. Med. Biol. Res. 2008, 41, 215–222. [CrossRef] [PubMed]
38. Ren, Y.; Zhang, J. Increased cardiorespiratory synchronization evoked by a breath controller based on heartbeat detection. BioMed.
Eng. OnLine 2019, 18, 61. [CrossRef] [PubMed]
39. Sola-Soler, J.; Cuadros, A.; Giraldo, B.F. Cardiorespiratory phase synchronization increases during certain mental stimuli in
healthy subjects. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 2018, 2018, 5298–5301. [CrossRef]
40. Schäfer, C.; Rosenblum, M.G.; Kurths, J.; Abel, H.H. Heartbeat synchronized with ventilation. Nature 1998, 392, 239–240.
[CrossRef]
41. Mizobuchi, A.; Osawa, K.; Tanaka, M.; Yumoto, A.; Saito, H.; Fuke, S. Detrended fluctuation analysis can detect the impairment
of heart rate regulation in patients with heart failure with preserved ejection fraction. J. Cardiol. 2021, 77, 72–78. [CrossRef]
42. Salo, T.M.; Sundell, J.; Knuuti, J.; Kemppainen, J.; Stolen, K.; Nuutila, P.; Mäkikallio, T.H.; Huikuri, H.V.; Airaksinen, K.E. Fractal
scaling properties of heart rate dynamics and myocardial efficiency in dilated cardiomyopathy. Clin. Res. Cardiol. 2009, 98,
725–730. [CrossRef]
43. Tsai, C.H.; Ma, H.P.; Lin, Y.T.; Hung, C.S.; Huang, S.H.; Chuang, B.L.; Lin, C.; Lo, M.T.; Peng, C.K.; Lin, Y.H. Usefulness of heart
rhythm complexity in heart failure detection and diagnosis. Sci. Rep. 2020, 10, 14916. [CrossRef] [PubMed]
44. Patel, V.N.; Pierce, B.R.; Bodapati, R.K.; Brown, D.L.; Ives, D.G.; Stein, P.K. Association of Holter-derived heart rate variability
parameters with the development of congestive heart failure in the cardiovascular health study. JACC Heart Fail. 2017, 5, 423–431.
[CrossRef] [PubMed]
45. DeMazumder, D.; Kass, D.A.; O’Rourke, B.; Tomaselli, G.F. Cardiac resynchronization therapy restores sympathovagal balance in
the failing heart by differential remodeling of cholinergic signaling. Circ. Res. 2015, 116, 1691–1699. [CrossRef] [PubMed]
46. Castiglioni, P.; Parati, G.; Di Rienzo, M.; Carabalona, R.; Cividjian, A.; Quintin, L. Scale exponents of blood pressure and heart
rate during autonomic blockade as assessed by detrended fluctuation analysis. J. Physiol. 2011, 589, 355–369. [CrossRef]
47. Platisa, M.M.; Gal, V. Influence of breathing frequency on short-term scaling exponent and spectral powers of RR interval series.
In Proceedings of the 6th Conference of the European Study Group on Cardiovascular Oscillations, Berlin, Germany, 12–14 April
2010; p. 74.
48. Perakakis, P.; Taylor, M.; Martinez-Nieto, E.; Revithi, I.; Vila, J. Breathing frequency bias in fractal analysis of heart rate variability.
Biol. Psychol. 2009, 82, 82–88. [CrossRef]
49. Kamath, C. A new approach to detect congestive heart failure using detrended fluctuation analysis of electrocardiogram signals.
J. Eng. Sci. Technol. 2015, 10, 145–159.
50. Platiša, M.M.; Radovanović, N.N.; Kalauzi, A.; Milašinović, G.; Pavlović, S.U. Differentiation of heart failure patients by the ratio
of the scaling exponents of cardiac interbeat intervals. Front. Physiol. 2019, 10, 570. [CrossRef]
51. Shin, D.G.; Lee, S.H.; Yi, S.H.; Yoo, C.S.; Hong, G.R.; Kim, U.; Park, J.S.; Kim, Y.J. Breakdown of the intermediate-term fractal
scaling exponent in sinus node dysfunction. New method for non-invasive evaluation of sinus node function. Circ. J. 2011, 75,
2775–2780. [CrossRef]
52. Peng, C.K.; Mietus, J.E.; Liu, Y.; Lee, C.; Hausdorff, J.M.; Stanley, H.E.; Goldberger, A.L.; Lipsitz, L.A. Quantifying fractal dynamics
of human respiration: Age and gender effects. Ann. Biomed. Eng. 2002, 30, 683–692. [CrossRef] [PubMed]
53. Raoufy, M.R.; Ghafari, T.; Darooei, R.; Nazari, M.; Mahdaviani, S.A.; Eslaminejad, A.R.; Almasnia, M.; Gharibzadeh, S.; Mani,
A.R.; Hajizadeh, S. Classification of asthma based on nonlinear analysis of breathing pattern. PLoS ONE 2016, 11, e0147976.
[CrossRef] [PubMed]