fphys-11-596567 January 15, 2021 Time: 13:10 # 1
published: 15 January 2021
University of South Australia, Australia
Federal University of Mato Grosso Do
MSB Medical School Berlin, Germany
This article was submitted to
a section of the journal
Frontiers in Physiology
Received: 19 August 2020
Accepted: 18 December 2020
Published: 15 January 2021
Rogers B, Giles D, Draper N,
Hoos O and Gronwald T (2021) A
New Detection Method Deﬁning
the Aerobic Threshold for Endurance
Exercise and Training Prescription
Based on Fractal Correlation
Properties of Heart Rate Variability.
Front. Physiol. 11:596567.
A New Detection Method Deﬁning
the Aerobic Threshold for Endurance
Exercise and Training Prescription
Based on Fractal Correlation
Properties of Heart Rate Variability
Bruce Rogers1*, David Giles2, Nick Draper3, Olaf Hoos4and Thomas Gronwald5
1College of Medicine, University of Central Florida, Orlando, FL, United States, 2Lattice Training Ltd., Chesterﬁeld,
United Kingdom, 3School of Health Sciences, College of Education, Health and Human Development, University
of Canterbury, Christchurch, New Zealand, 4Center for Sports and Physical Education, Julius Maximilians University
of Wuerzburg, Wuerzburg, Germany, 5Department of Performance, Neuroscience, Therapy and Health, Faculty of Health
Sciences, MSH Medical School Hamburg, University of Applied Sciences and Medical University, Hamburg, Germany
The short-term scaling exponent alpha1 of detrended ﬂuctuation analysis (DFA a1), a
nonlinear index of heart rate variability (HRV) based on fractal correlation properties, has
been shown to steadily change with increasing exercise intensity. To date, no study has
speciﬁcally examined using the behavior of this index as a method for deﬁning a low
intensity exercise zone. The aim of this report is to compare both oxygen intake (VO2)
and heart rate (HR) reached at the ﬁrst ventilatory threshold (VT1), a well-established
delimiter of low intensity exercise, to those derived from a predeﬁned DFA a1 transitional
value. Gas exchange and HRV data were obtained from 15 participants during an
incremental treadmill run. Comparison of both VO2and HR reached at VT1 deﬁned by
gas exchange (VT1 GAS) was made to those parameters derived from analysis of DFA
a1 reaching a value of 0.75 (HRVT). Based on Bland Altman analysis, linear regression,
intraclass correlation (ICC) and ttesting, there was strong agreement between VT1
GAS and HRVT as measured by both HR and VO2. Mean VT1 GAS was reached at
39.8 ml/kg/min with a HR of 152 bpm compared to mean HRVT which was reached at
40.1 ml/kg/min with a HR of 154 bpm. Strong linear relationships were seen between
test modalities, with Pearson’s rvalues of 0.99 (p<0.001) and.97 (p<0.001) for VO2
and HR comparisons, respectively. Intraclass correlation between VT1 GAS and HRVT
was 0.99 for VO2and 0.96 for HR. In addition, comparison of VT1 GAS and HRVT
showed no differences by ttesting, also supporting the method validity. In conclusion, it
appears that reaching a DFA a1 value of 0.75 on an incremental treadmill test is closely
associated with crossing the ﬁrst ventilatory threshold. As training intensity below the
ﬁrst ventilatory threshold is felt to have great importance for endurance sport, utilization
of DFA a1 activity may provide guidance for a valid low training zone.
Keywords: detrended ﬂuctuation analysis, ventilatory threshold, aerobic threshold, intensity distribution, intensity
zones, endurance exercise, endurance training, polarized training
Frontiers in Physiology | www.frontiersin.org 1January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 2
Rogers et al. DFA a1 Aerobic Threshold
Training zone identiﬁcation is part of the foundation for exercise
intensity distribution study and implementation (Stöggl and
Sperlich, 2019. Traditionally, the upper limit of the low intensity
range (zone 1 in a 3 zone model) for intensity distribution
for endurance exercise and training prescription has been
represented by the ﬁrst ventilatory (VT1) or lactate threshold
(LT1) (Seiler and Kjerland, 2006;Esteve-Lanao et al., 2007;Mann
et al., 2013;Pallarés et al., 2016). Although there may be diﬀerent
schools of thought on what type of distribution is “optimal”
(polarized vs. pyramidal or threshold) both models are deﬁned
by having the major portion of training in zone 1. In addition,
several training approaches for endurance athletes recommend
spending large amounts of exercise time in a low intensity zone
(Muñoz et al., 2014;Stöggl and Sperlich, 2014, 2019;Bourgois
et al., 2019;Casado et al., 2019). Gold standard methods to obtain
VT1 or LT1 revolve around either formal gas exchange testing
or invasive blood lactate sampling. These procedures can be
costly, require special test equipment, trained operators, ongoing
calibration and veriﬁcation. Even if these methods are utilized,
there is disagreement on their accuracy as both visual (Yeh
et al., 1983;Meyer et al., 1996) and automated (Ekkekakis et al.,
2008) gas exchange analysis can be subject to substantial error.
In addition, LT1 assessment can vary depending on the chosen
concept of determination (Newell et al., 2007;Faude et al., 2009;
Jamnick et al., 2018). Training guided by erroneous intensity
targets could lead to potential adverse consequences such as
prolonged cardiac parasympathetic recovery (Seiler and Kjerland,
2006;Stanley et al., 2013), central and muscular fatigue (Noakes
et al., 2005;Venhorst et al., 2018), glycogen depletion (Beneke
et al., 2011), and gastrointestinal barrier disruption (van Wijck
et al., 2012). In view of the diﬃculties involved in gas exchange
analysis, lactate test availability, invasiveness, and accuracy, a
search for alternate methods of identifying the limits of low
intensity exercise seem worthwhile.
Cardiac interbeat interval variation, commonly referred to
as heart rate variability (HRV), has been extensively studied
in both resting states (Shaﬀer and Ginsberg, 2017) as well as
during dynamic exercise (Hottenrott and Hoos, 2017;Michael
et al., 2017). Certain HRV indexes have been observed to change
as exercise intensity rises, potentially providing information
regarding an individual’s physiologic status (Tulppo et al., 1996;
Casties et al., 2006;Sandercock and Brodie, 2006;Karapetian
et al., 2008;Michael et al., 2017;Gronwald et al., 2018,
2019a,b,c). It has also been shown that several of the examined
HRV indexes also change during lower intensities (Tulppo
et al., 1996;Sandercock and Brodie, 2006;Karapetian et al.,
2008;Botek et al., 2010;Michael et al., 2017) making them
potentially suitable for zone 1 delineation. However, despite some
initial interest, widespread usage for the speciﬁc purpose of
low intensity training limitation has not occurred. Frequency-
domain parameters such as high frequency (HF) power have
been noted to be unreliable in a sizable fraction of individuals
with up to 20% of subjects not having identiﬁable breakpoints
(Cottin et al., 2007). Time domain measures such as SDNN
were found to closely relate with VT1 but little follow-up or
veriﬁcation has been done (Karapetian et al., 2008). The SD1
is another index that has been examined during exercise. It is
based on a Poincare plot of each RR interval graphed against
the preceding interval and is related to short term trends in
RR patterns often assigned to nonlinear indexes (Shaﬀer and
Ginsberg, 2017), although it is mathematically equivalent to
another time domain index (Ciccone et al., 2017). While showing
potential as a low intensity marker in some earlier studies
(Tulppo et al., 1996) other evidence indicates that SD1 was
already suppressed in young athletes at the ﬁrst tested work rate
of 60% VO2MAX making it less useful for zone 1 delineation
(Blasco-Lafarga et al., 2017).
One nonlinear index, the short-term scaling exponent
alpha1 based on Detrended Fluctuation Analysis (DFA a1),
has generated interest as both an indicator of autonomic
nervous system regulation as well as an overall marker of
organismic demands Gronwald and Hoos, 2020). Originally,
Peng et al. (1995) developed this method to measure scale-
invariant behavior; this involved the evaluation of trends of all
sizes in the presence or absence of fractal correlation properties
in a heart rate (HR) time series (Yeh et al., 2010). Thus, the DFA
method allows for the quantiﬁcation of the degree of correlation
and fractal scale of a HRV signal resulting in dimensionless
measures. The short-term scaling exponent DFA a1 is based
on the fractal dynamics (self-similarity) of the cardiac beat-to-
beat pattern and provides insights into correlation properties
of HR time series caused by physiological processes (Peng
et al., 1995). DFA a1 values indicate time series correlation
properties with approximately 1.5 indicating a strongly correlated
pattern and ≤0.5 for anti-correlated pattern with random
behavior; approximately 1.0 signiﬁes a mix of uncorrelated
and maximally correlated signal components (represents a
balance between complete unpredictability (randomness) and
predictability (strong correlations), also associated with fractal
(self-similar) behavior (Platisa and Gal, 2008). Larger values
of DFA a1 represent a smoother time series and smaller
values of DFA a1 represent coarser ones (Peng et al., 1995;
Goldberger et al., 2002). Within this framework, DFA a1
has been shown to decline as work rate rises, starting from
strongly correlated patterns (value of 1.5) at rates well below
the ﬁrst ventilatory threshold (VT1), transitioning (values
of 1.0–0.5) through values representing uncorrelated, less
complex white noise behavior at moderate to high work
rates, then ﬁnally showing anti-correlated behavior at the
highest intensities (values of <0.5) (Gronwald et al., 2019c;
Gronwald and Hoos, 2020). Given this relationship, there
may be an opportunity to assist athletes in delineating
intensity training zones by observing the change in DFA
a1 with increasing exercise intensity (Gronwald et al., 2020;
The purpose of this report is to validate a predeﬁned DFA
a1 value of 0.75 with the exercise intensity at VT1 obtained
during an incremental treadmill run to exhaustion. This is to
be done by a direct comparison of the VT1 intensity based on
both absolute VO2and HR obtained during gas exchange with
the same measures derived from analysis of DFA a1 behavior.
If it can be shown that a predeﬁned “boundary” value in the
Frontiers in Physiology | www.frontiersin.org 2January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 3
Rogers et al. DFA a1 Aerobic Threshold
DFA a1 index occurs near the VT1, this could establish a basis
for further research exploiting a non-linear autonomic nervous
system related marker in prospective exercise and training
MATERIALS AND METHODS
Seventeen male volunteers aged 19–52, without previous medical
history, current medications or physical issues were tested.
A background questionnaire regarding medical history was
reviewed along with information of the potential testing risks
then institutionally approved consent was given. Approval
for the study was granted by the University of Derby,
United Kingdom (LSREC_1415_02) and conformed to the
principles of the Declaration of Helsinki. Participants did not
consume caﬀeine, alcohol or any stimulant for the 24 h before
testing. Background data for each subject included, age, body
weight, and training volume in hours per week (Table 1). All
testing was done in the afternoon and at least 3 h post meal.
No exercise was performed the day prior to the test. Two
participants with a high degree of cardiac ectopy (frequent
atrial premature beats and atrial trigeminy) during testing were
excluded from analysis.
Participants performed an incremental VO2MAX test on a
motorized treadmill (Woodway, Birmingham, United Kingdom).
The treadmill was set for the Bruce protocol with increases
in speed and inclination from 2.7 km/h at ten percent grade,
increasing by 1.3 km/h and two percent grade every 3 min until
volitional exhaustion. A fan was used for cooling.
TABLE 1 | Demographic data of all included participants (n= 15) with
Subject number Age (years) BW (Kg) HT (cm) TV (h/wk)
1 19 82 182 3–6
2 19 82 176 3–6
3 20 82 190 3–6
4 23 77 180 >6
5 24 69 171 3–6
6 24 65 165 >6
7 24 76 186 3–6
8 25 78 171 >6
9 26 69 169 >6
10 30 92 189 1–3
11 30 73 175 >6
12 32 65 161 1–3
13 36 75 182 >6
14 50 94 178 3–6
15 52 71 171 1–3
Mean (SD) 29 (±10) 77 (±9) 176 (±9) -
BW, Body weight; TV, Training volume. Mean (±standard deviation, SD) in last row.
Gas Exchange Testing and Calculation of
the First Ventilatory Threshold
Gas exchange kinetics were recorded continuously using a
breath-to-breath metabolic cart (Metalyzer 3B; Cortex Biophysik
GmbH Germany). In addition, a Polar H7 (Polar Electro Oy,
Kempele, Finland) was wirelessly paired to the Metalyzer cart for
the purpose of HR recording concurrent with gas exchange data.
VO2, VCO2, PetO2, PetCO2, Ve/VO2, Ve/VCO2, and HR were
imported into Microsoft Excel 365 for analysis. The native gas
exchange analysis feature of the Metalyzer was not used due to
the unreliability of many automated VT1 calculations (Ekkekakis
et al., 2008). Graphing of the above parameters were done to
derive VT1, VO2MAX, and VO2vs. time. No averaging was done
for either gas exchange parameters or HR. Inspection of the VO2
over time relationship was done to determine any signiﬁcant
plateau of the VO2curve for estimation of VO2MAX and VO2
linearity. If a signiﬁcant plateau was found, compensation for
calculating both VO2MAX and the VO2over time equation was
done. To reduce the chance of failure to identify the VT1 by gas
exchange (VT1 GAS) based on a single method, evaluation was
done according to the triple detection method consisting of V
slope, Ve/VO2, and excess CO2from Gaskill et al. (2001) as well
as the PetO2nadir from Binder et al. (2008). Based on the quality
and consistency of the plots, the excess CO2method was chosen
to be used for all participants and reviewed independently by two
investigators (Figure 1A). VO2was plotted over the elapsed time
of the incremental test to produce a linear regression equation.
VO2at the time of VT1 was based on linear regression from the
VO2over time relationship.
RR Measurements and Calculation of
DFA a1 Derived Threshold
A 3-lead ECG (MP36; Biopac Systems Ltd.) with a sampling
rate of 1,000 Hz was used to record the subject’s ECG/RR
times series. Biopac ﬁlter settings were set to 0.05 Hz high-
pass ﬁlter and 150 Hz low-pass ﬁlter. Electrodes were placed
in the CM5 distribution after appropriate skin cleansing and
shaving if necessary. Sample data from the MP36 was saved
as .acq ﬁles. ECG ﬁles for each subject were imported into
Kubios 3.3.2 (Biosignal Analysis and Medical Imaging Group,
Department of Physics, University of Kuopio, Kuopio, Finland).
Kubios preprocessing settings were at the default values including
the RR detrending method which was kept at “Smoothn priors”
(Lambda = 500, Tarvainen et al., 2014). For DFA a1 estimation,
the root mean square ﬂuctuation of the integrated and detrended
data is measured in observation windows of diﬀerent sizes. The
data are then plotted against the size of the window on a log-log
scale. The scaling exponent represents the slope of the line, which
relates (log) ﬂuctuation to (log) window size (Mendonca et al.,
2010). DFA a1 window width was set to 4 ≤n≤16 beats.
For the detection of a HRV derived threshold, a DFA a1 value
of 0.75 was chosen based on this being the midpoint between
a fractal behavior of the HR time series of 1.0 (seen with very
light exercise) and an uncorrelated value of 0.5 which represents
white noise, random behavior (seen with high intensity exercise).
A value of 0.75 has also been used as a cut-oﬀ value for survival
Frontiers in Physiology | www.frontiersin.org 3January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 4
Rogers et al. DFA a1 Aerobic Threshold
FIGURE 1 | (A) Usage of “Excess CO2” technique to determine VT1 GAS.
The intersection of the baseline and ﬁrst rise in excess CO2corresponds to
the time when VT1 occurs; (B) DFA a1 plotted over time, the area of linear
drop of DFA a1 from about 1.0 to 0.5 is used to determine the VO2(from the
VO2vs. time relation) at HRVT; (C) DFA a1 plotted against HR, the area of
linear drop of DFA a1 from 1.0 to 0.5 is used to determine the HR at HRVT. All
data taken from Subject #2.
curves and mortality rate assessment during resting conditions
(Huikuri et al., 2000).
The following procedure was used to indicate at what level of
running intensity (as VO2or HR) the DFA a1 would cross a value
of 0.75: DFA a1 was calculated from the incremental exercise test
RR series using 2 min time windows with a recalculation every
5 s throughout the test. Two minute time windowing was chosen
based on the reasoning of Chen et al. (2002). The rolling time
window measurement was used to better delineate rapid changes
in the DFA a1 index over the course of the test. Each DFA a1 value
is based on the RR series 1 min pre and 1 min post the designated
time stamp. For example, at a time of 10 min into the testing, the
DFA a1 is calculated from the 2 min window starting from minute
9 and ending at minute 11 and labeled as the DFA a1 at 10 min.
Based on a rolling time recalculation every 5 s, the next data point
would occur at 10:05 min (start 9:05 min and end 11:05 min).
Plotting of DFA a1 vs. time was then performed. Inspection
of the DFA a1 relationship with time generally showed a reverse
sigmoidal curve with a stable area above 1.0 at low work rates,
a rapid, near linear drop reaching below 0.5 at higher intensity,
then ﬂattening without major change. A linear regression was
done on the subset of data consisting of the rapid near linear
decline from values near 1.0 (correlated) to approximately 0.5
(uncorrelated). The time of DFA a1 reaching 0.75 was calculated
based on the linear regression equation from that straight section
(Figure 1B). The time of DFA a1 reaching 0.75 was then
converted to VO2using the VO2vs. time relation, resulting in the
VO2at which DFA a1 equaled 0.75 (HRVT). A similar analysis
was done for the HR reached at a DFA a1 of 0.75. First, ECG data
from each 2 min rolling window was used to plot the average HR
and DFA a1. The HR at which DFA a1 equaled 0.75 was found
using the same technique as above, a linear regression through
the rapid change section of DFA a1 values of 1.0 to below 0.5,
with a subsequent equation for HR and DFA a1 (Figure 1C).
Using a ﬁxed variable of DFA a1 equals 0.75, the resulting HR was
obtained. The HR at DFA a1 0.75 (based on ECG data) was then
compared to the HR at VT1 GAS obtained from the metabolic
cart data (based on the Polar H7).
Visual inspection of the entire test recording was done to
determine sample quality, noise, arrhythmia, and missing beat
artifact. As mentioned above, two participants with a high degree
of atrial ectopy were excluded from analysis. The RR series
of the included participants was then corrected by the Kubios
“automatic method” and exported as text ﬁles for further analysis.
Percent artifact reported refers those occurring during the linear
regression segment (DFA a1 1.0 to near 0.5).
Statistical analysis was performed for the main variables, VO2at
VT1 derived from gas exchange testing, VO2at DFA a1 0.75,
HR at VT1 obtained from gas exchange testing and average
HR at DFA a1 0.75. Standard statistical methods were used for
the calculation of means and standard deviations (SD). Normal
distribution of data was checked by Shapiro–Wilk’s test. The
agreement against the Gold Standard VT1 GAS was assessed
using intraclass correlation coeﬃcient (ICC), linear regression,
Pearson’s rcorrelation coeﬃcient, standard error of estimate
(SEE), coeﬃcient of determination (R2) and Bland Altman plots
with limits of agreement (Bland and Altman, 1999). The size of
Pearson’s rcorrelations evaluated as follows; 0.3 ≤r<0.5 low;
0.6 ≤r<0.8 moderate and r≥0.8 high (Chan, 2003). The paired
Frontiers in Physiology | www.frontiersin.org 4January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 5
Rogers et al. DFA a1 Aerobic Threshold
t-test was used for comparison of VT1 GAS vs. HRVT for both
VO2and HR parameters. For all tests, the statistical signiﬁcance
was accepted as p≤0.05. Cohen’s dwas used to denote eﬀect sizes
(small eﬀect = 0.2, medium eﬀect = 0.5, large eﬀect = 0.8; Cohen,
1988). Analysis was performed using Microsoft Excel 365 with
Real Statistics Resource Pack software (Release 6.8).
Gas Exchange Testing
Individual gas exchange results are presented in Table 2. Both the
VO2MAX as well as the percentage of VO2MAX and HR at VT1
GAS varied considerably among participants. VO2MAX ranged
between 41 and 74 ml/kg/min. VT1 GAS was reached between 61
and 86% of the VO2MAX and at HRs between 108 and 183 bpm.
RR Interval Quality
The percentage of artifacts was calculated based on the
Kubios automatic correction method for each subject’s test
data. Since only a portion of the entire treadmill test was
used for the linear interpolation of DFA a1, the artifact
percentage listed refers to that section only. Artifact percentage
for the linear plotted data series was between 0 and 3%,
all consisting of atrial premature complexes (Table 2).
There were no missed beats due to noise interference or
loss of electrode contact. The two participants originally
excluded from analysis had signiﬁcant ectopy, leading to an
uninterpretable DFA a1 pattern.
Comparison of VT1 GAS vs. HRVT
The average VT1 GAS was 39.8 ml/kg/min (±8.9) compared
to 40.1 ml/kg/min (±8.6) obtained by HRVT. The average HR
at VT1 GAS was 152 bpm (±21) compared to 154 bpm (±20)
obtained by HRVT. Strong linear relationships were seen between
test modalities, with Pearson’s rvalues of 0.99 (p<0.001) and.97
(p<0.001) for VO2and HR comparisons respectively (Figure 2).
Intraclass correlation between VT1 GAS and HRVT was 0.99 for
VO2and 0.96 for HR. The comparison of VT1 GAS and HRVT
showed no diﬀerences (VO2:p= 0.347, d= 0.030; HR: p= 0.191,
d= 0.091). Bland Altman analysis for VT1 GAS vs. HRVT for
VO2(Figure 3) showed a mean diﬀerence of −0.33 ml/kg/min
(±1.3) with upper and lower limits of 2.2 and −2.9 ml/kg/min.
Bland Altman analysis for VT1 GAS vs. HRVT for HR (Figure 3)
showed a mean diﬀerence of −1.9 bpm (±5) with upper and
lower limits of 8 and −12 bpm.
This study explored whether values of the nonlinear HRV index,
DFA a1, pass through a deﬁned transitional zone at workloads
near VT1 during an incremental treadmill test. Since many
prior reports have shown DFA a1 to decline during incremental
exercise (Hautala et al., 2003;Casties et al., 2006;Platisa
et al., 2008;Karavirta et al., 2009;Blasco-Lafarga et al., 2017;
Gronwald et al., 2019c), our result showing a similar occurrence
is not unanticipated. However, none have attempted to directly
examine the possibility that the DFA a1 index has a distinct
value at the VT1 work rate. Since many of the prior studies
looking at DFA a1 response to incremental exercise intensity have
used cycling as the exercise modality, it is also reassuring to see
analogous results with treadmill running, adding validity to the
behavior of this index during other endurance exercise types.
The inclusion of a wide range of subject ages, body weights and
ﬁtness abilities, lends strength to the application of our results
TABLE 2 | Comparison of VT1 GAS and HRVT with measures of VO2and HR.
Subject number VO2MAX VT1 GAS VO2VT1 GAS VO2HRVT VO2VT1 GAS HR HRVT HR Artifacts (%)
(ml/kg/min) (%MAX) (ml/kg/min) (ml/kg/min) (bpm) (bpm)
1 58 77 45.2 46.1 167 170 0
2 57 75 42.8 45.1 169 175 0.5
3 47 70 32.7 31.6 178 175 0
4 71 86 61.2 61.2 155 156 0
5 64 68 43.6 43.0 143 137 0
6 54 74 40.1 38.1 165 163 0
7 47 76 35.8 37.2 164 171 2
8 54 70 37.8 37.9 137 135 0
9 72 69 49.3 49.2 183 184 0
10 46 61 27.7 29.2 108 122 0
11 74 65 48.1 46.4 151 148 3
12 49 66 32.0 33.3 154 160 1
13 57 66 37.6 39.4 154 159 1
14 41 70 28.6 28.8 139 136 0
15 54 64 34.5 35.5 118 122 1
Mean (SD) 56 (±10) 70 (±6) 39.8 (±8.9) 40.1 (±8.6) 152 (±21) 154 (±20) 0.6 (±0.9)
VT1 GAS, ﬁrst ventilatory threshold; HRVT, DFA a1 derived threshold; HR, Heart rate; VO2MAX , VT1 GAS percent of VO2MAX and Artifacts percentage. Mean (±standard
deviation, SD) in last row.
Frontiers in Physiology | www.frontiersin.org 5January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 6
Rogers et al. DFA a1 Aerobic Threshold
FIGURE 2 | Regression plots for all subject data. (A) Values of VT1 GAS vs. HRVT for VO2;(B) Values of VT1 GAS vs. HRVT for HR. Bisection lines in light gray. SEE,
standard error of estimate; R2, coefﬁcient of determination.
to the general population and its application in diﬀerent ﬁelds of
physical exercise and training.
In a recent perspective review (Gronwald et al., 2020),
identiﬁcation of a low intensity exercise zone based on DFA a1
for the purposes of endurance exercise and training prescription
was discussed. The mechanism underlying DFA a1 decline with
exercise is felt to be related to autonomic balance and a complex
interaction of the two main branches, namely parasympathetic
withdrawal, sympathetic intensiﬁcation as well as other factors
(Gronwald and Hoos, 2020). Since VT1 is usually seen at a point
of signiﬁcant parasympathetic withdrawal (Tulppo et al., 1996;
Sales et al., 2019), leveraging HRV parameters that reliably reﬂect
this occurrence can be of use during endurance exercise and
training. Our methodology to determine HRVT utilized the rapid
decline of DFA a1 from 1.0 to below 0.5, seen during progressive
exercise intensity. The results presented here appear to indicate
Frontiers in Physiology | www.frontiersin.org 6January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 7
Rogers et al. DFA a1 Aerobic Threshold
FIGURE 3 | Bland Altman Plot of VT1 GAS vs. HRVT for all participants. (A) Values of VT1 GAS vs. HRVT for VO2;(B) Values of VT1 Gas vs. HRVT for HR. Center
line in each plot represents the mean difference between each paired value, the top and bottom lines are 1.96 standard deviations from the mean difference.
that VT1 is reached at a midpoint between a fractal behavior of
DFA a1 and a pattern of uncorrelated white noise with random
behavior, corresponding to a DFA a1 of approximately 0.75.
Bland Altman analysis with limits of agreement showed minimal
diﬀerence between VT1 GAS and HRVT looking at either VO2
or HR measurements. Correlation coeﬃcients and ICC were high
for both VO2and HR based comparisons.
Although the DFA a1 value of 0.75 was chosen theoretically, a
brief review of prior investigation is supportive of this ﬁgure. In a
study of young men performing a cycling ramp test, an average
DFA a1 of 0.49 was associated with a lactate measurement of
2.49, indicating that LT1 had already been exceeded (Gronwald
et al., 2019c). Other cycling ramp studies in men of diﬀerent
ﬁtness levels seemed to indicate that DFA a1 crossed the value
of 0.75 at about 73–78% of VO2MAX (Hautala et al., 2003;
Hottenrott and Hoos, 2017), within the approximate realm of
VT1 for many individuals (Gaskill et al., 2001;Pallarés et al.,
2016). An examination of the DFA a1 response to incremental
cycling exercise in teenage males (Blasco-Lafarga et al., 2017)
showed an approximate crossing of the 0.75 value at an average
Frontiers in Physiology | www.frontiersin.org 7January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 8
Rogers et al. DFA a1 Aerobic Threshold
intensity near 65% of maximum, also near published ranges of
VT1 (50–65% of VO2MAX) in that age group (Runacres et al.,
2019). In the current study, there appeared to be little bias in
the VO2or HR associated with HRVT. Although there was
some variability in VT1 GAS vs. HRVT parameters, for the
most part, associations were similar to that of other comparisons
of threshold approaches such as blood lactate or ventilatory
parameters (Pallarés et al., 2016), or assessment of gas exchange
techniques for VT1 determination (Gaskill et al., 2001). Several
participants had relatively high heart rate at the VT1 (Azevedo
et al., 2011) of which we do not have an explanation. A strength of
this study is that of RR interval quality. Direct ECG visualization
was done and a research grade device with a high sample rate
was used. No missed or lost beats were seen, and the only
artifact type present was APC aberrancy. In view of reports
indicating substantial bias in nonlinear HRV indexes with artifact
presence (Giles and Draper, 2018;Rincon Soler et al., 2018), a
weakening of DFA a1 derived VT1 accuracy could occur with
higher artifact occurrence.
A signiﬁcant advantage of DFA a1 over other HRV indexes
for the determination of a low intensity threshold revolves
around the nature of testing. Other HRV metrics proposed
to identify VT1 such as SDNN (Karapetian et al., 2008), HF
power (Cottin et al., 2007) or SD1 (Tulppo et al., 1996) require
testing into high intensity zones since they rely on curve
interpretation that displays a demonstrable nadir. With DFA
a1, once the VT1 boundary area is reached, little additional
increase in exercise intensity should be required. The potential
beneﬁt of utilizing a ﬁxed DFA a1 value as the VT1 delineation
marker is especially attractive in populations unable or ill
advised to enter high intensity regions. In addition, for athletes
evaluating low intensity training limits, avoidance of exercise
ramps to volitional failure may help avert undue stress in a
polarized training model.
Limitations and Future Direction
Given the issues with both availability and accuracy of gas
exchange or blood lactate testing in determining VT1 for training
zone purposes, an alternate modality that employs relatively
simple wearable technology seems attractive. However, while
DFA a1 monitoring may be a promising approach, several
questions need to be addressed. Although this study was done
with a wide range of subject age and ﬁtness characteristics, no
female participants were tested. If the DFA a1 index behavior is
to be considered as a zone 1 delimiter for the general population,
further investigation using female subjects is mandatory. Another
area of concern is the transfer of the DFA a1 0.75 breakpoint
obtained during incremental testing to that of one found during
constant load exercise, including moderate length intervals
(5 min). No data is available comparing DFA a1 behavior during
an incremental ramp to constant load exercise (Gronwald and
Hoos, 2020), making automatic transfer of zone boundaries
unclear. Whether the index will remain stable for even longer
exercise intervals (>60 min) performed below VT1 intensity is
another open question as well as day to day repeatability. Another
interesting subject to explore is the impact of athlete overtraining
on DFA a1 behavior and VT1 prediction accuracy during
exercise. Baumert et al. (2006) did show changes in DFA related
scaling behavior after intense training, which may provide both a
potential source of HRVT bias and an opportunity to screen for
overtraining states. Although it seems that ramp protocol slope
has minimal eﬀect on the VT1 gas assessment (Weston et al.,
2002;Boone and Bourgois, 2012), the analogous assumption
needs to be shown in terms of HRVT thresholds. Another area
for investigation is whether DFA a1 cut oﬀ values are equivalent
between chest belt and research grade ECG recordings. Although
in this study, the RR intervals were recorded with a research
grade ECG device, it may be possible to reproduce similar
results with chest belt ECG recordings. In that regard, two
major questions need to be addressed. One is that of exercise
associated missed beat artifact with possible faulty interpolation
strategies by interpreting software, creating potential bias in
the calculated DFA a1 values. As mentioned above, several
reports have questioned the degree of bias of nonlinear HRV
indexes if artifacts are present in the RR series (Giles and
Draper, 2018;Rincon Soler et al., 2018;Stapelberg et al., 2018).
Artifacts may be of diﬀerent types such as missed beats or
aberrancy. In the current report, no missed beats were seen,
and only relatively rare atrial premature complexes were noted.
However, two participants exhibited frequent APC aberrancy,
had uninterpretable DFA a1 curves and were excluded from
group analysis. Given the relative low numbers of participants, no
deﬁnitive conclusion can be reached regarding artifact bias, but
further investigation into eﬀects of missed beats and aberrancy
on the use of DFA a1 to delineate zone 1 transition is needed.
Second is the question of DFA a1 value precision obtained by
diverse monitoring devices possessing diﬀerent sample rates and
prepossessing strategies. Device sample rates have been shown to
variably alter DFA a1 values at rest (Voss et al., 1996;Tapanainen
et al., 1999;Singh et al., 2015) but may have more signiﬁcant
eﬀects during exercise. Although no recent caﬀeine use was
noted by history, we have no information on prior long term
intake patterns which could aﬀect autonomic balance on abrupt
discontinuation (La Monica et al., 2018). Finally, it may be
possible to answer many of these questions by “repurposing”
prior work already done. For instance, a study by Boullosa
et al. (2014) assessed the changes in DFA a1 before and after
a typical incremental treadmill ramp to exhaustion. A look
back at previously acquired RR recordings during the active
ramp portion using the methods discussed here may be a way
to rapidly acquire needed information about DFA a1 behavior
during dynamic exercise.
DFA a1, an index of fractal dynamics and correlations properties
of the heart rate time series, was noted to decline during an
incremental treadmill run test to exhaustion. The area of most
rapid change of this index occurred near the ﬁrst ventilatory
threshold. The point of DFA a1 reaching a value of 0.75 during
the incremental treadmill test was directly associated with the
ﬁrst ventilatory threshold as measured by heart rate and gas
exchange VO2. As training intensity below the ﬁrst ventilatory
threshold is felt to have great importance for exercise and
training prescription in endurance sport, utilization of DFA a1
Frontiers in Physiology | www.frontiersin.org 8January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 9
Rogers et al. DFA a1 Aerobic Threshold
activity may provide guidance for a valid low training zone
boundary without the need for gas exchange or blood lactate
testing. Further study of DFA a1 behavior in female participants,
during constant load intervals, index stability over long periods
of time and across diverse recording devices is recommended.
If investigation into these matters remain consistent with the
results presented here, obtaining a low intensity zone boundary
by automated analysis of a training session via an appropriate
wearable device may be possible.
DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be
made available by the authors, without undue reservation.
The studies involving human participants were reviewed
and approved by University of Derby, United Kingdom
(LSREC_1415_02). The patients/participants provided their
written informed consent to participate in this study.
BR and TG conceived the study. DG and ND performed the
physiologic testing. BR wrote the ﬁrst draft of the article. BR and
TG performed the data analysis. All authors revised it critically
for important intellectual content, ﬁnal approval of the version to
be published, and accountability for all aspects of the work.
Azevedo, L. F., Perlingeiro, P. S., Brum, P. C., Braga, A. M., Negrão, C. E., and
de Matos, L. D. (2011). Exercise intensity optimization for men with high
cardiorespiratory ﬁtness. J. Sports Sci. 29, 555–561. doi: 10.1080/02640414.
Baumert, M., Brechtel, L., Lock, J., and Voss, A. (2006). Changes in heart rate
variability of athletes during a training camp. Biomed. Tech. 51, 201–204. doi:
Beneke, R., Leithäuser, R. M., and Ochentel, O. (2011). Blood lactate diagnostics in
exercise testing and training. Int. J. Sports Physiol. Perf. 6, 8–24. doi: 10.1123/
Binder, R. K., Wonisch, M., Corra, U., Cohen-Solal, A., Vanhees, L., Saner, H., et al.
(2008). Methodological approach to the ﬁrst and second lactate threshold in
incremental cardiopulmonary exercise testing. Eur. J. Cardiovasc. Prev. Rehab.
15, 726–734. doi: 10.1097/hjr.0b013e328304fed4
Bland, J. M., and Altman, D. G. (1999). Measuring agreement in method
comparison studies. Stat. Methods Med. Res. 8, 135–160. doi: 10.1191/
Blasco-Lafarga, C., Camarena, B., and Mateo-March, M. (2017). Cardiovascular
and Autonomic Responses to a Maximal Exercise Test in Elite Youngsters. Int.
J. Sports Med. 38, 666–674. doi: 10.1055/s-0043-110680
Boone, J., and Bourgois, J. (2012). The oxygen uptake response to incremental ramp
exercise: methodogical and physiological issues. Sports Med. 1 42, 511–526.
doi: 10.2165/11599690-000000000- 00000
Botek, M., Stejskal, P., Krejci, J., Jakubec, A., and Gaba, A. (2010). Vagal threshold
determination. Eﬀect of age and gender. Int. J. Sports Med. 31, 768–772. doi:
Boullosa, D. A., Barros, E. S., del Rosso, S., Nakamura, F. Y., and Leicht, A. S. (2014).
Reliability of heart rate measures during walking before and after running
maximal eﬀorts. Int. J. Sports Med. 35, 999–1005. doi: 10.1055/s-0034-13
Bourgois, J. G., Bourgois, G., and Boone, J. (2019). Perspectives and Determinants
for Training-Intensity Distribution in Elite Endurance Athletes. Int.
J. Sports Physiol. Perform. 14, 1151–1156. doi: 10.1123/ijspp.2018-
Casado, A., Hanley, B., Santos-Concejero, J., and Ruiz-Pérez, L. M. (2019). World-
Class Long-Distance Running Performances Are Best Predicted by Volume
of Easy Runs and Deliberate Practice of Short-Interval and Tempo Runs.
J. Strength. Cond. Res. doi: 10.1519/JSC.0000000000003176 [Online ahead of
Casties, J. F., Mottet, D., and Le Gallais, D. (2006). Non-linear analyses of heart rate
variability during heavy exercise and recovery in cyclists. Int. J. Sports Med. 27,
780–785. doi: 10.1055/s-2005-872968
Chan, Y. H. (2003). Biostatistics 104: correlational analysis. Singapore Med. J. 44,
Chen, Z., Ivanov, P. C., Hu, K., and Stanley, H. E. (2002). Eﬀect of nonstationarities
on detrended ﬂuctuation analysis. Phys. Rev. E 65:041107.
Ciccone, A. B., Siedlik, J. A., Wecht, J. M., Deckert, J. A., Nguyen, N. D., and
Weir, J. P. (2017). Reminder: RMSSD and SD1 are identical heart rate variability
metrics. Muscle Nerve. 56, 674–678. doi: 10.1002/mus.25573
Cohen, J. (1988). Statistical Power Analysis for the Behavioural Sciences. Hillsdale:
Cottin, F., Médigue, C., Lopes, P., Leprêtre, P. M., Heubert, R., and Billat, V.
(2007). Ventilatory thresholds assessment from heart rate variability during
an incremental exhaustive running test. Int. J. Sports Med. 28, 287–294. doi:
Ekkekakis, P., Lind, E., Hall, E. E., and Petruzzello, S. J. (2008). Do regression-based
computer algorithms for determining the ventilatory threshold agree? J. Sports
Sci. 26, 967–976. doi: 10.1080/02640410801910269
Esteve-Lanao, J., Foster, C., Seiler, S., and Lucia, A. (2007). Impact of training
intensity distribution on performance in endurance athletes. J. Strength Cond.
Res. 21, 943–949. doi: 10.1519/00124278-200708000-00048
Faude, O., Kindermann, W., and Meyer, T. (2009). Lactate threshold concepts.
Sports Med. 39, 469–490. doi: 10.2165/00007256-200939060-00003
Gaskill, S. E., Ruby, B. C., Walker, A. J., Sanchez, O. A., Serfass, R. C., and Leon,
A. S. (2001). Validity and reliability of combining three methods to determine
ventilatory threshold. Med. Sci. Sports. Exerc. 33, 1841–1848. doi: 10.1097/
Giles, D. A., and Draper, N. (2018). Heart Rate Variability During Exercise:
A Comparison of Artefact Correction Methods. J. Strength Cond. Res. 32,
726–735. doi: 10.1519/jsc.0000000000001800
Goldberger, A. L., Amaral, L. A., Hausdorﬀ, J. M., Ivanov, P. C., Peng, C. K., and
Stanley, H. E. (2002). Fractal dynamics in physiology: Alterations with disease
and aging. PNAS 99, 2466–2472. doi: 10.1073/pnas.012579499
Gronwald, T., and Hoos, O. (2020). Correlation properties of heart rate variability
during endurance exercise: A systematic review. Ann. Noninvas. Electrocardiol.
Gronwald, T., Hoos, O., and Hottenrott, K. (2019a). Eﬀects of Acute Normobaric
Hypoxia on Non-linear Dynamics of Cardiac Autonomic Activity During
Constant Workload Cycling Exercise. Front. Physiol. 10:999. doi: 10.3389/fphys.
Gronwald, T., Hoos, O., and Hottenrott, K. (2019b). Eﬀects of a Short-Term
Cycling Interval Session and Active Recovery on Non-Linear Dynamics of
Cardiac Autonomic Activity in Endurance Trained Cyclists. J. Clin. Med. 6:8.
Gronwald, T., Hoos, O., Ludyga, S., and Hottenrott, K. (2019c). Non-linear
dynamics of heart rate variability during incremental cycling exercise. Res.
Sports Med. 27, 88–98. doi: 10.1080/15438627.2018.1502182
Gronwald, T., Ludyga, S., Hoos, O., and Hottenrott, K. (2018). Non-linear
dynamics of cardiac autonomic activity during cycling exercise with varied
cadence. Hum. Mov. Sci. 60, 225–233. doi: 10.1016/j.humov.2018.06.013
Gronwald, T., Rogers, B., and Hoos, O. (2020). Fractal correlation properties of
heart rate variability: A new biomarker for intensity distribution in endurance
exercise and training prescription? Front. Physiol. 10:3389. doi: 10.3389/fphys.
Hautala, A. J., Mäkikallio, T. H., Seppänen, T., Huikuri, H. V., and Tulppo, M. P.
(2003). Short-term correlation properties of R-R interval dynamics at diﬀerent
exercise intensity levels. Clin. Physiol. Funct. Imag. 23, 215–223. doi: 10.1046/j.
Hottenrott, K., and Hoos, O. (2017). “Heart rate variability analysis in exercise
physiology,” in ECG time series analysis: Engineering to medicine, eds H. Jelinek,
A. Khandoker, and D. Cornforth (London, UK: CRC Press), 245–257.
Frontiers in Physiology | www.frontiersin.org 9January 2021 | Volume 11 | Article 596567
fphys-11-596567 January 15, 2021 Time: 13:10 # 10
Rogers et al. DFA a1 Aerobic Threshold
Huikuri, H. V., Mäkikallio, T. H., Peng, C. K., Goldberger, A. L., Hintze, U., and
Møller, M. (2000). Fractal correlation properties of RR interval dynamics and
mortality in patients with depressed left ventricular function after an acute
myocardial infarction. Circulation 101, 47–53. doi: 10.1161/01.cir.101.1.47
Jamnick, N. A., Botella, J., Pyne, D. B., and Bishop, D. J. (2018). Manipulating
graded exercise test variables aﬀects the validity of the lactate threshold and
VO2peak. PLoS One 13:e0199794. doi: 10.1371/journal.pone.0199794
Karapetian, G. K., Engels, H. J., and Gretebeck, R. J. (2008). Use of heart rate
variability to estimate LT and VT. Int. J. Sports Med. 29, 652–657. doi: 10.1055/
Karavirta, L., Tulppo, M. P., Laaksonen, D. E., Nyman, K., Laukkanen, R. T.,
Kinnunen, H., et al. (2009). Heart rate dynamics after combined endurance
and strength training in older men. Med. Sci. Sports Exerc. 41, 1436–1443.
La Monica, M. B., Fukuda, D. H., Wang, R., Gonzalez, A. M., Wells, A. J., Hoﬀman,
J. R., et al. (2018). Maintenance of Vagal Tone with Time-Release Caﬀeine, But
Vagal Withdrawal During Placebo in Caﬀeine-Habituated Men. J. Caﬀ. Adenos.
Res. 8, 59–64. doi: 10.1089/caﬀ.2017.0039
Mann, T., Lamberts, R. P., and Lambert, M. I. (2013). Methods of prescribing
relative exercise intensity: physiological and practical considerations. Sports
Med. 43, 613–625. doi: 10.1007/s40279-013-0045-x
Mendonca, G. V., Heﬀernan, K. S., Rossow, L., Guerra, M., Pereira, F. D., and
Fernhall, B. (2010). Sex diﬀerences in linear and nonlinear heart rate variability
during early recovery from supramaximal exercise. Appl. Physiol. Nutr. Metab.
35, 439–446. doi: 10.1139/h10-028
Meyer, K., Hajric, R., Westbrook, S., Samek, L., Lehmann, M., S chwaibold,M., et al.
(1996). Ventilatory and lactate threshold determinations in healthy normals
and cardiac patients: methodological problems. Eur. J. Appl. Physiol. Occupat.
Physiol. 72, 387–393. doi: 10.1007/bf00242266
Michael, S., Graham, K. S., and Oam Davis, G. M. (2017). Cardiac Autonomic
Responses during Exercise and Post-exercise Recovery Using Heart Rate
Variability and Systolic Time Intervals-A Review. Front. Physiol. 8:301. doi:
Muñoz, I., Seiler, S., Bautista, J., España, J., Larumbe, E., and Esteve-Lanao, J.
(2014). Does polarized training improve performance in recreational runners?
Int. J. Sports Physiol. Perform. 9, 265–272. doi: 10.1123/ijspp.2012- 0350
Newell, J., Higgins, D., Madden, N., Cruickshank, J., Einbeck, J., McMillan, K., et al.
(2007). Software for calculating blood lactate endurance markers. J. Sports Sci.
25, 1403–1409. doi: 10.1080/02640410601128922
Noakes, T. D., Gibson, A. S. C., and Lambert, E. V. (2005). From catastrophe to
complexity: a novel model of integrative central neural regulation of eﬀort and
fatigue during exercise in humans: summary and conclusions. Br. J. Sports Med.
39, 120–124. doi: 10.1136/bjsm.2003.010330
Pallarés, J. G., Morán-Navarro, R., Ortega, J. F., Fernández-Elías, V. E., and Mora-
Rodriguez, R. (2016). Validity and Reliability of Ventilatory and Blood Lactate
Thresholds in Well-Trained Cyclists. PLoS One. 11:e0163389. doi: 10.1371/
Peng, C. K., Havlin, S., Stanley, H. E., and Goldberger, A. L. (1995). Quantiﬁcation
of scaling exponents and crossover phenomena in nonstationary heartbeat time
series. Chaos 5, 82–87. doi: 10.1063/1.166141
Platisa, M. M., and Gal, V. (2008). Correlation properties of heartbeat dynamics.
Eur. Biophys. J. 37, 1247–1252. doi: 10.1007/s00249-007- 0254-z
Platisa, M. M., Mazic, S., Nestorovic, Z., and Gal, V. (2008). Complexity of
heartbeat interval series in young healthy trained and untrained men. Physiol.
Meas. 29, 439–450. doi: 10.1088/0967-3334/29/4/002
Rincon Soler, A. I., Silva, L. E. V., Fazan, R. Jr., and Murta, L. O. Jr. (2018). The
impact of artefact correction methods of RR series on heart rate variability
parameters. J. Appl. Physiol. 124, 646–652. doi: 10.1152/japplphysiol.00927.
Rogers, B. (2020). A personalized low intensity exercise prescription based on an
index of non linear heart rate variability: a case report. J. Athl. Enhanc. 9:1.
Runacres, A., Mackintosh, K. A., and McNarry, M. A. (2019). The eﬀect of
constant-intensity endurance training and high-intensity interval training on
aerobic and anaerobic parameters in youth. J. Sports Sci. 37, 2492–2498. doi:
Sales, M. M., Sousa, C. V., da Silva Aguiar, S., Knechtle, B., Nikolaidis, P. T.,
Alves, P. M., et al. (2019). An integrative perspective of the anaerobic threshold.
Physiol. Behav. 205, 29–32. doi: 10.1016/j.physbeh.2017.12.015
Sandercock, G. R., and Brodie, D. A. (2006). The use of heart rate variability
measures to assess autonomic control during exercise. Scand. J. Med. Sci. Sports
16, 302–313. doi: 10.1111/j.1600-0838.2006.00556.x
Seiler, K. S., and Kjerland, G. Ø (2006). Quantifying training intensity distribution
in elite endurance athletes: is there evidence for an “optimal” distribution?
Scand. J. Med. Sci. Sports 16, 49–56. doi: 10.1111/j.1600-0838.2004.00418.x
Shaﬀer, F., and Ginsberg, J. P. (2017). An Overview of Heart Rate Variability
Metrics and Norms. Front. Publ. Health. 5:258. doi: 10.3389/fpubh.2017.00258
Singh, M., Singh, M., and Banga, V. (2015). “Optimal ECG Sampling Rate for Non-
Linear Heart Rate Variability,” in IJCA Proceedings on International Conference
on Advancements in Engineering and Technology ICAET (Korea: ICAET),
Stanley, J., Peake, J. M., and Buchheit, M. (2013). Cardiac parasympathetic
reactivation following exercise: implications for training prescription. Sports
Med. 43, 1259–1277. doi: 10.1007/s40279-013-0083-4
Stapelberg, N. J. C., Neumann, D. L., Shum, D. H. K., McConnell, H., and
Hamilton-Craig, I. (2018). The sensitivity of 38 heart rate variability measures
to the addition of artefact in human and artiﬁcial 24-hr cardiac recordings. Ann.
Noninvas. Electrocardiol. 23:e12483.
Stöggl, T., and Sperlich, B. (2014). Polarized training has greater impact on key
endurance variables than threshold, high intensity, or high volume training.
Front. Physiol. 5:33. doi: 10.3389/fphys.2014.00033
Stöggl, T. L., and Sperlich, B. (2019). Editorial: Training Intensity, Volume and
Recovery Distribution Among Elite and Recreational Endurance Athletes.
Front. Physiol. 10:592. doi: 10.3389/fphys.2019.00592
Tapanainen, J. M., Seppänen, T., Laukkanen, R., Loimaala, A., and Huikuri,
H. V. (1999). Signiﬁcance of the Accuracy of RR Interval Detection for the
Analysis of New Dynamic Measures of Heart Rate Variability. Anna. Noninvas.
Electrocardiol. 4, 10–17. doi: 10.1111/j.1542-474x.1999.tb00359.x
Tarvainen, M. P., Niskanen, J. P., Lipponen, J. A., Ranta-Aho, P. O., and
Karjalainen, P. A. (2014). Kubios HRV–heart rate variability analysis software.
Comput. Methods Progr. Biomed. 113, 210–220.
Tulppo, M. P., Mäkikallio, T. H., Takala, T. E., Seppänen, T., and Huikuri,
H. V. (1996). Quantitative beat-to-beat analysis of heart rate dynamics during
exercise. Am. J. Physiol. 271, H244–H252.
van Wijck, K., Lenaerts, K., Grootjans, J., Wijnands, K. A., Poeze, M., Van Loon,
L. J., et al. (2012). Physiology and pathophysiology of splanchnic hypoperfusion
and intestinal injury during exercise: strategies for evaluation and prevention.
Am. J. Physiol. 303, G155–G168.
Venhorst, A., Micklewright, D., and Noakes, T. D. (2018). Towards a three-
dimensional framework of centrally regulated and goal-directed exercise
behaviour: a narrative review. Br. J. Sports Med. 52, 957–966. doi: 10.1136/
Voss, A., Wessel, N., Sander, A., Malberg, H., and Dietz, R. (1996). Requirements
on sampling rate in Holter systems for analysis of heart rate variability. Clin. Sci.
91(Suppl.), 120–121. doi: 10.1042/cs0910120supp
Weston, S. B., Gray, A. B., Schneider, D. A., and Gass, G. C. (2002). Eﬀect of ramp
slope on ventilation thresholds and VO2peak in male cyclists. Int. J. Sports Med.
23, 22–27. doi: 10.1055/s-2002-19267
Yeh, R. G., Chen, G. Y., Shieh, J. S., and Kuo, C. D. (2010). Parameter investigation
of detrended ﬂuctuation analysis for short-term human heart rate variability. J.
Med. Biol. Eng. 30, 277–282. doi: 10.5405/jmbe.30.5.02
Yeh, M. P., Gardner, R. M., Adams, T. D., Yanowitz, F. G., and Crapo, R. O. (1983).
“Anaerobic threshold”: problems of determination and validation. J. Appl.
Physiol. 55, 1178–1186. doi: 10.1152/jappl.1922.214.171.1248
Conﬂict of Interest: DG was employed by company Lattice Training.
The remaining authors declare that the research was conducted in the absence of
any commercial or ﬁnancial relationships that could be construed as a potential
conﬂict of interest.
Copyright © 2021 Rogers, Giles, Draper, Hoos and Gronwald. This is an open-access
article distributed under the terms of the Creative Commons Attribution License
(CC BY). The use, distribution or reproduction in other forums is permitted, provided
the original author(s) and the copyright owner(s) are credited and that the original
publication in this journal is cited, in accordance with accepted academicpractice. No
use, distribution or reproduction is permitted which does not comply with theseterms.
Frontiers in Physiology | www.frontiersin.org 10 January 2021 | Volume 11 | Article 596567