Heart rate variability (HRV) analysis quantifies fluctuations in the duration of interbeat intervals (29). Ideally, HRV analysis is performed on RR intervals that contain only pure sinus beats (normal to normal intervals; NN). However, both technical and biological artefacts occur in the data (15–17). If artefacts are left unedited, they introduce bias and present a significant problem for the interpretation of calculated time, frequency, and nonlinear parameters (23); as such, preprocessing is necessary. Preprocessing of RR interval data involves the editing of artefacts by deletion, interpolation, or filtering. However, editing methods have distinct effects on the RR interval time series and HRV results. If appropriate corrections are performed, the presence of artefacts should have a less dramatic influence on HRV data (27).
Consideration of the validity of RR interval data recorded using consumer heart rate monitors (HRMs) before analysis is essential, especially because they are increasingly being used in research (11,31,35). Their portability has enabled the recording of RR data in situations where it was not previously possible and not conceivable with laboratory-based electrocardiograms (ECG) or even ambulatory ECGs (21,22). Finally, and most significantly, previous research has demonstrated the validity of HRMs at rest, finding good agreement with a simultaneously recorded ECG with small but acceptable levels of variation (e.g., Polar V800 (17), RS800 (25,33), S810 (15,16,32,35)). The validity of recorded RR intervals holds true as long as artefacts are corrected before analysis. While it is not possible to determine the source of artefacts in HRM signal; it is believed that they result because of a decrease in contact between the skin and the electrode, causing a reduction in the amplitude of the detected R-wave (16,17).
Artefacts require correction even at rest when their occurrence is minimal (15–17). Simply discarding erroneous sections of data, especially sections that involve more movement and potentially a greater number of artefacts, leads to bias (9). However, the methods that are used for the correction of artefacts in recent literature vary considerably. Although several studies have used various interpolation techniques (8,12), others simply deleted offending intervals (20) or, most frequently, do not mention correction at all (13,14,31). Equally, research that has specifically examined the validity of HRMs has used several methods (4,32). As such, there is a need for standard practice in the collection and processing of RR interval data recorded using HRMs in research as, currently, there are no standard criteria. Therefore, this study aimed to examine current methods for error correction, along with the extent of alterations in artefact occurrence during exercise, in comparison with simultaneously recorded ECG.
Experimental Approach to the Problem
RR interval data were simultaneously recorded using the Polar V800 and a 3-lead ECG at rest and during incremental exercise. Artefacts were (a) identified following the proposed guidelines of Giles, Draper and Neil (17) developed from Gamelin, Berthoin, and Bosquet (16); (b) artefacts in the HRM RR interval time series were corrected by deletion, interpolation, and using Kubios HRV; (c) time, frequency, and nonlinear HRV parameters were calculated for ECG RR intervals, uncorrected HRM interval, and for each correction methods; (d) agreement was assessed using percentage bias and effect size (ES), intraclass correlation coefficients (ICC), and Bland-Altman limits of agreement (LoA).
Eighteen male volunteers (mean ± SD age 27.6 ± 9.9 years, range = 18–39 years; height 177.4 ± 9.2 cm; mass 77.0 ± 7.9 kg) agreed to participate in the study. No participant was known to be taking medication or have any cardiovascular or respiratory diseases or illnesses that may have influenced the procedures performed. Participants were informed of the benefits and risks of the investigation before completing and signing institutionally approved informed consent and medical health questionnaires. Approval for the study was granted by the University of Derby, UK [LSREC_1415_16] and conformed to the principles of the Declaration of Helsinki.
RR Interval Recording
Participants were asked to abstain from caffeine-containing food and drink for 24 hours before the test. Participant's skin was cleaned (shaved if necessary) and prepared for the attachment of the ECG electrodes. The electrodes were placed in a CM5 configuration (right fifth interspace, manubrium, and left fifth interspace) ensuring that they did not interfere with the fit of the HRM strap (Polar H7; Polar Electro Oy, Kempele, Finland). The electrode belt was dampened and placed, following Polar's guidelines, tightly but comfortably just below the chest muscles.
Resting measurements were conducted in 2 positions, supine and, following an active lying-to-standing orthostatic challenge, standing. Participants were kept in silence, without moving, and with limited audible and visual stimulation. Recordings lasted for 10 minutes in the supine position and 7 minutes in standing position. No attempt was made to control participant's respiratory frequency or tidal volume (24). Environmental conditions were maintained at 20.6 ± 1.0° C and 50% relative humidity.
Participants performed an incremental V̇o2max test, using a modified Bruce protocol, on a motor-driven treadmill (Woodway, Birmingham, United Kingdom). The treadmill was programmed for increases in angle of inclination and speed from 2.7 km·h−1 at 10% grade, increasing by 1.3 km·h−1 (0.6 mph) and 2% grade every 3 minutes until volitional exhaustion. Oxygen kinetics were recorded continuously using a breath-to-breath ergospirometry system (Metalyzer 3B; Cortex Biophysik GmbH Germany).
RR interval data were recorded simultaneously using a V800 Polar HRM with a Polar H7 chest strap and a 3-lead ECG (MP36; Biopac Systems Ltd.), at a sampling frequency of 1,000 Hz for both devices. R-wave peaks from the ECG were detected automatically using a custom peak detection algorithm in Matlab (Mathworks, Cambridge). The raw ECG traces and detected R-waves were manually assessed to ensure that they had been correctly detected. Ectopic beats were noted, but not corrected at that stage of analysis.
RR Interval Data Handling
Data were saved as RR interval data files, with intervals in milliseconds (ms). For the Polar HRM raw, unfiltered RR data were exported from the Polar Flow web service as a space delimited .txt file. Both the ECG and HRM raw RR signal start points were manually matched before further analysis. Interval data were divided into the following sections: Supine, Standing, < 40% V̇o2max, 40–60% V̇o2max, 60–80% V̇o2max, 80–100% V̇o2max, and active recovery (walking).
Artefact Handling, Correction Procedure, and Identification of Artefacts
The ECG and Polar V800 interval time series were compared side by side to identify artefacts and ectopic beats greater than 20 milliseconds (Figure 1). Ectopic beats are not easily differentiated from artefacts in HRM RR signals; ectopic beats present identically to the artefacts listed in Table 1 (usually as a type 4 or type 1) and should be corrected as such. The methods proposed for the correction of artefacts and ectopic beats are identical to those used previously in replacement of ectopic beats in ECG signals (23,26). For the purpose of this article, nonsinus beats (0.0015% RR intervals) were replaced during analysis in both signals with interpolated data from adjacent RR intervals. Although this does not represent real-world use, it ensured that bias from the ectopic beats was not introduced. Before correction, discrepancies between the 2 signals were identified and synchronicity maintained with the insertion of a 0 millisecond interval.
Categorization of Artefacts
After visual identification, discrepancies were assigned to one of the 6 types of artefacts given in Table 1. The artefacts identified are based on the research of Giles, Draper and Neil (17) developed from Gamelin, Berthoin, and Bosquet (16). The frequency of the occurrence of each of the artefacts was recorded.
Methods for Correction
Once identified, artefacts in the RR time series were corrected. Previously, Gamelin, Baquet, Berthoin, and Bosquet (15) corrected all artefacts detected (T1-T5); however, this does not represent typical use of the HRM, as, without the utilization of a simultaneous ECG recording it is not possible to detect T1 artefacts that are less than approximately a 50% increase in length over the previous interval. Equally, T6-a is undetectable without a simultaneous recording. Giles, Draper and Neil (17) proposed the correction of unidentifiable artefacts (short T1 and T6-b), while still correcting those identifiable without a simultaneous ECG recording (T1-T5 and T6-b) following the guidelines given in Table 1. The exception for the application of correction methods was T5, which were simply summed in all correction methods. The following correction methods were selected, based on their applicability to RR interval signals recorded by HRMs and the ease of their application to the RR interval data. When selecting methods for the identification and correction of artefacts, consideration was given to (a) the ease of applying the identification and correction methods to data, ensuring that they do not require preparatory software or specific programming knowledge; (b) do not need a simultaneous ECG for their identification; (c) are applicable to RR interval recordings completed at rest and during exercise; and (d), finally, are easy to interpret.
The methods used for artefact correction were (a) Uncorrected, no correction applied to any intervals; data were left as recorded. (b) Deletion, erroneous RR interval(s) were simply deleted from the time series. Deletion can have a significant effect on HRV parameters because of changes in the length of the signal, particularly in short-term recordings and frequency-domain parameters (23,26). Deletion introduces step-like shapes into the RR interval time series resulting in changes in variability as well as decreasing the length of the signal, producing false low frequency (LF) and high frequency (HF) components (26). Interpolation: Interpolation methods, in contrast to the deletion, replace erroneous nonnormal RR intervals with interpolated intervals. Critically, interpolation allows for the length of the recording to remain the same, mitigating the issue of reduced signal length. (c) Degree Zero (average), substitution of artefact(s) with a mean value that is calculated from surrounding RR intervals. On longer sections of artefacts, degree zero interpolation results in the same averaged value over a whole segment, resulting in a flat shape, introducing false trends, and increasing LF and very low frequency (VLF) power (5,26). (d) Degree One (linear), a straight line is fitted over the irregular intervals to obtain new values. As with degree zero, on longer sections of artefacts, slope-like shapes occur, introducing false trends and potentially increasing LF and VLF components (5,26). (e) Cubic, cubic interpolation uses 4 datum points to compute the polynomial; there are no constraints on the derivatives. Cubic interpolation does not result in flat sections of data. However, as nonlinear analysis is concerned with the complexity and irregularity of heartbeat series, the introduction of a potentially falsely correlated signal is of concern, particular if there is a significant number of erroneous intervals (23,26). (f) Spline, cubic spline interpolation. Smooth values are estimated through a number of datum points by fitting a third-degree polynomial. Cubic spline interpolation computes a third-order polynomial from only 2 datum points with the additional constraint that the first and second derivatives at the interpolation points are continuous. As with cubic interpolation, spline interpolation may also introduce false correlations into the signal. (g) Kubios HRV software (Version 2.2), Kubios software provides options for the detection and correction of ectopic beats and artefacts. Ectopic beat–free RR interval data were entered into Kubios; the appropriate correction level was chosen by visual inspection of the interval tachogram (the threshold for detecting artefact beats), to ensure that the correction removed the artefacts but did not distort normal RR intervals. When the correction was applied, detected artefact beats were automatically replaced using cubic spline interpolation (28).
Calculation of Heart Rate Variability Parameters and Statistical Analysis
Following the analysis of the Polar RR trace for artefacts and the replacement of ectopic, erroneous and noisy complexes, the RR interval data were considered normal, and thus described as NN data. All HRV analysis was performed with the freely available Kubios HRV software (Version 2.2; 28).
Heart Rate Variability Parameters
For the calculation of HRV, all intervals from each section were used for ECG and corrected HRM recordings. These selected segments were analyzed for time, frequency domain, and nonlinear components. Not all parameters are relevant to exercise or the length of the recording but are presented for clarity and completeness. (a) Time-domain analysis concerns the statistical representation of the variation in NN intervals within the sample (19). A number of parameters were calculated: SDNN is the standard deviation of the NN intervals, RMSSD, the root mean square of the successive difference of intervals, and pNN50%, the number of successive intervals that differ by more than 50 ms, expressed as a % of the total (19). (b) Frequency-domain analysis allows for the identification of sympathetic and parasympathetic contributions of HRV. The nonparametric power spectral density analysis provides basic information on how power, and therefore the variance, distributes as a function of frequency using a fast Fourier transformation. A fast Fourier transformation allows the analysis of the components of the power spectrum density to be quantified into different frequency bands (1). Three spectral components were calculated: VLF (0.00–0.04 Hz), LF (0.04–0.15 Hz), and HF (0.15–0.40 Hz). Additionally, normalized LF and HF power were calculated (as a percentage of the sum of LF and HF power) and the ratio LF:HF power. (c) Nonlinear analysis, given the complex control systems of the heart, it is reasonable to assume nonlinear mechanisms are involved in the genesis of HRV; nonlinear analysis of NN intervals describes the chaotic nature of the signal (28). The values SD1 and SD2 were calculated from a Poincaré plot of the data, by fitting an ellipse oriented according to the line of identity and computing the standard deviation of the points perpendicular (SD1) to and along the line of identity (SD2) (7). The analysis comprised of Sample Entropy was also calculated, measuring the complexity of the NN series, low entropy arises from extremely regular time series, higher values reflect more complexity, and the highest values are typical for stochastic datasets (34).
Descriptive statistics were first calculated for all variables; all values are reported as mean ± SD. Normal distribution and homogeneity of variance were assessed by visual inspection of the frequency histogram and with either a Kolmogorov-Smirnov (RR intervals) or Shapiro-Wilk test (HRV parameters) depending on the number of samples. Homoscedasticity was determined through the analysis of the plot of the standardized residuals. A 1-way repeated-measures analysis of variance (ANOVA) was used to assess changes in the frequency of artefacts. Percentage bias was calculated between each of the correction methods and the ECG RR intervals . The magnitude of the difference of the RR intervals and the HRV parameters was calculated by determining the ES which represents the mean difference over the standard deviation of the difference (30); the difference was considered small when ES ≤ 0.2, moderate when ES ≤ 0.5, and great when ES > 0.8 (10). Relative reliability was assessed for all variables by calculating the ICC (36), model 3.1 was used (3). Bland-Altman 95% LoA were calculated for all RR and HRV parameters (6). If heteroscedasticity was present in any HRV data, it was log transformed before the calculation of the LoA. The level for accepting the statistical significance of tests was set at p ≤ 0.05 for all analysis. All data were analyzed using SPSS (Version 22; Chicago, IL, USA).
Alterations in select time, frequency, and nonlinear HRV parameters are provided in Table 2 for supine and standing resting measurements, exercise and active recovery.
The number of RR intervals detected (59,545 total) is shown in Table 3, along with the type and frequency of artefacts identified. Type 4 made up 96.8% of detected artefacts, type 6a 1.4%, type 3 1%, and the remaining artefacts <1%. No type 2 artefacts were detected. The frequency of the occurrence of artefacts increased significantly with the rise in the exercise intensity (ANOVA, p > 0.005).
Along with the increased number of artefacts, uncorrected HRM RR intervals saw increased percentage bias, LoA, and ES with an increase in exercise intensity. Except for supine and standing recordings (0.09 and 0.05%, respectively), the number of artefacts resulted in moderate-to-large ES and increased bias (Table 4), demonstrating the need for correction in all recordings, except those with less than ∼0.1% artefacts. All correction methods resulted in small percentage bias and ES for intervals recorded during resting and <60% V̇o2max (Table 4). Linear interpolation produced corrected intervals with the lowest bias and ES. However, values of RMSSD, LF:HF ratio, SD1, and SampEn at 60 + V̇o2max exercise intensities showed large bias and ES, and increased LoA and reduced ICC, regardless of correction (See Tables in Supplemental Digital Content, https://links.lww.com/JSCR/A33, for raw ICC and LoA for each of the correction methods at each exercise intensity).
This study aimed to establish differences in the occurrence of artefacts in RR intervals during exercise and to determine the validity of methods for their correction. Briefly, artefacts increased relative to exercise intensity, to a peak of 4.46% during recordings made at 80–100% V̇o2max. Artefact correction was necessary, with large percentage bias and ES of HRV parameters in all but supine and standing recordings; correction resulted in reduced bias and ES for resting and <60% V̇o2max recordings with all methods. Linear interpolation consistently produced HRM RR intervals with the lowest percentage bias and ES. However, caution should be given to the interpretation of RMSSD, LF:HF ratio, SD1, and SampEn at high (60% + V̇o2max) exercise intensities, as, even when correction methods were applied, large amounts of bias were still present.
It is not possible to determine the source of the detected artefacts they may result because of increased activity of chest muscles (respiration and arm movement) and repetitive vertical oscillations causing interference and loss, or decrease, in contact between the skin and the chest strap electrode, reducing the amplitude of the detected R-wave (16,17). Exercise that does not introduce either of these factors is unlikely to dramatically influence the number of artefacts detected. As most recent HRM validation studies were conducted either at rest or during an exercise that did not involve upper body movement (e.g., cycle ergometry), the observed occurrence of artefacts has previously been very low, at less than 1% (12,20,31). The criterion for the identification of artefacts also varies considerably among studies and, as such, they may also be significantly underreported.
The increased occurrence of artefacts and the large bias and ES demonstrated that correction was necessary before HRM RR interval data could be considered NN (23). Only supine and standing recordings, with an artefact rate of <0.1% produced acceptable uncorrected HRV data. Except for T1 and T6-a, artefacts should be recognizable and correctable without the use of a simultaneous ECG recording. The potential for artefacts may be reduced by following the manufacturer's guidelines for fitting and positioning of the strap. External factors may also interfere with the transmission of the Bluetooth signal from the chest strap to the watch, particularly cross-body and wireless network interference (18). As the position of the watch was kept constant for all tests and participants and there were no other operating wireless devices in the laboratory, it is unlikely that these factors affected the results of this study.
The deletion of artefacts is the simplest of correction techniques; however, as HRV spectral analysis is sensitive to changes in the signal length, discontinuity of the signal, and loss of samples the current study does not support the use of deletion (29). Previous research that has used deletion to correct RR intervals should be treated with caution. Interpolation of artefacts maintains the length of the RR interval time series, overcoming the main issue of deletion. Linear interpolation produced the smallest bias and ES in most, but not all cases. In line with the results of this study, previous authors have found interpolation to increase the reliability of RR interval data (2). However, inserting linearly interpolated intervals can create flat sections with little or no variability, and it is advised that large sections of ectopic beats should not be edited using these techniques (23). From the results of this study, it is not possible to state the number or frequency of artefacts that may be considered acceptable. However, caution should be given to the correction of RR intervals with sections that have more than 4 concurrent artefacts, or that contain a notable number within a small area of data. Future research could examine a threshold for the inclusion (or exclusion) of excessively noisy signals to determine whether to correct or discard the RR interval data.
The manual correction of RR intervals is time consuming; Kubios HRV software provides options for the automatic detection and correction of ectopic beats and artefacts. However, despite Kubios appearing to accurately detect most artefacts, the results showed larger bias and ES compared with both the ECG and the manually corrected cubic spline interpolation. Both correction methods used the same interpolation calculation in Matlab. Figure 2 shows a short section of data with a single type 4 artefact present; it is apparent that although Kubios software does accurately identify the artefact (Figure 2a), the erroneous beat is only replaced with a single interval, rather than the required 2 (Figure 2b). Although this does mean that the signal length remains constant, the number of intervals is one, or more, less than if correctly manually. It would appear that the reduced validity occurs because of this error. Until this issue has been addressed, the automatic correction methods used within Kubios HRV software (Version 2.2) are not recommended for RR interval data recorded using HRMs, where accuracy is important.
Heart rate monitors are regularly used for the recording of RR intervals and the calculation of HRV in exercise-related studies (12,20,31). However, both technical and biological artefacts occur in the RR interval data, which if left unedited, introduce bias and present a significant problem for the interpretation of calculated time, frequency, and nonlinear HRV parameters (23). The occurrence of artefacts in data requires correction in all but recordings with >0.1% of artefacts. The impact of exercise on the number of artefacts detected is likely to be related to the increase in heart rate and movement in the chest, that cause artefacts in the RR interval time series to increase relative to exercise intensity. As the results of this study demonstrate, artefact correction is necessary for RR intervals obtained from HRMs. Correction of artefacts with a simple linear interpolation reduced bias and ES and increased ICC, in most but not all cases: caution should be given to RMSSD, LF:HF ratio, SD1, and SampEn at high (60 + V̇o2max) exercise intensities. Finally, the methods used in this study are recommended for future studies using HRMs in sport and exercise research, both at rest and during exercise.
Methods used for the correction of ectopic and erroneous signals vary considerably. It is recommended that researchers using HRMs to record RR intervals state how the occurrence of artefacts was minimized, how they were identified and corrected. The following guidelines are proposed to increase transparency and minimize bias in RR interval recordings made using HRMs:
The fit and conductance of the chest strap should be maximized, by choosing a correctly sized strap appropriate for the participant, ensuring that the electrodes are wet, and the chest strap is stable. Minimize environmental sources of interference by checking the correct pairing of the HRM and the chest strap, and avoid other wireless networks. Position the watch (if not on the wrist) in front of the participant.
Where possible, select sections of RR intervals that are artefact free. If artefact-free sections are not achievable, artefacts should be identified following the guidelines set out in this article. The percentage of artefacts identified should be stated in the methods, along with the inclusion/exclusion criteria. Caution should be given to sections of data that contain more than 4 concurrent artefacts, or more than 4 artefacts in close proximity.
Identified artefacts must be corrected before HRV analysis. The means of artefact correction should be clearly stated. The findings of this study support the use of linear interpolation.
No funding was received for this study. The authors have no competing interests to declare. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of this study do not constitute endorsement of any products by the authors or the NSCA.
1. Achten J, Jeukendrup AE. Heart rate monitoring. Sports Med 33: 517–538, 2003.
2. Albrecht P, Cohen RJ. Estimation of Heart Rate Power Spectrum Bands from Real-world Data: Dealing With Ectopic Beats and Noisy Data. Presented at IEEE Computers in Cardiology Conference; September 25–28, 1988; Washington, DC.
3. Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Med 26: 217–238, 1998.
4. Barbosa R, da Silva NT, de Azevedo FM, Pastre CM, Vanderlei LCM. Comparison of Polar® RS800G3™ heart rate monitor with Polar® S810i™ and electrocardiogram to obtain the series of RR intervals and analysis of heart rate variability at rest. Clin Physiol Funct Imaging 36: 112–117, 2014.
5. Birkett C, Kienzle M, Myers G. Interpolation over ectopic beats increases low frequency power in heart rate variability spectra. Presented at Computers in Cardiology Conference; September 23, 1991; Venice, Italy.
6. Bland JM, Altman D. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 327: 307–310, 1986.
7. Brennan M, Palaniswami M, Kamen P. Do existing measures of Poincare plot geometry reflect nonlinear features of heart rate variability? IEEE Trans Biomed Eng 48: 1342–1347, 2001.
8. Buchheit M, Solano R, Millet GP. Heart-rate deflection point and the second heart-rate variability threshold during running exercise in trained boys. Pediatr Exerc Sci 19: 192–204, 2007.
9. Citi L, Brown EN, Barbieri R. A real-time automated point-process method for the detection and correction of erroneous and ectopic heartbeats. IEEE Trans Biomed Eng 59: 2828–2837, 2012.
10. Cohen J. Statistical Power Analysis for the Behavioral Sciences. Oxford, United Kingdom: Academic press, 2013.
11. de Lima DC, Ribeiro HS, Cristina R, Oliveira M, de Vasconcelos Generoso S, Lima AS, Correia MITD. Functional status and heart rate variability in end-stage liver disease patients: Association with nutritional status. Nutrition 31: 971–974, 2015.
12. De Vito G, Galloway S, Nimmo MA, Maas P, McMurray JJ. Effects of central sympathetic inhibition on heart rate variability during steady-state exercise in healthy humans. Clin Physiol Funct Imaging 22: 32–38, 2002.
13. Duarte TS, Júnior AJF, Rezende EC, Pertence LC, Carlos J, Marins B, de Lima JRP. Association between heart rate variability threshold and post-exercise vagal reactivation. J Exerc Physiol Online 17: 58–65, 2014.
14. Fronchetti L, Nakamura FY, De-Oliveira FR, Lima-Silva AE, Lima J. Effects of high-intensity interval training on heart rate variability during exercise. J Exerc Physiol Online 10: 1–9, 2007.
15. Gamelin F, Baquet G, Berthoin S, Bosquet L. Validity of the polar S810 to measure RR intervals in children. Int J Sports Med 29: 134–138, 2008.
16. Gamelin FX, Berthoin S, Bosquet L. Validity of the polar S810 heart rate monitor to measure RR intervals at rest. Med Sci Sports Exerc 38: 887–893, 2006.
17. Giles D, Draper N, Neil W. Validation of the polar V800 heart rate monitor to measure RR intervals at rest. Eur J Appl Physiol 116: 563–571, 2016.
18. Kara A, Bertoni HL. Blockage/Shadowing and Polarization Measurements at 2.45 ghz for Interference Evaluation Between Bluetooth and IEEE 802.11 WLAN. Presented at IEEE Antennas and Propagation Society International Symposium: 2001; July 8–13, 2001; Boston, MA.
19. Karim N, Hasan JA, Ali SS. Heart rate variability–a review. J Basic Appl Sci 7: 71–77, 2011.
20. Lewis MJ, Kingsley M, Short A, Simpson K. Influence of high-frequency bandwidth on heart rate variability analysis during physical exercise. Biomed Signal Process Control 2: 34–39, 2007.
21. Mateo M, Blasco-Lafarga C, Martínez-Navarro I, Guzmán JF, Zabala M. Heart rate variability and pre-competitive anxiety in BMX discipline. Eur J Appl Physiol 112: 113–123, 2012.
22. Morales J, Garcia V, García-Massó X, Salvá P, Escobar R. The use of heart rate variability in assessing precompetitive stress in high-standard judo athletes. Int J Sports Med 34: 144–151, 2013.
23. Peltola MA. Role of editing of R–R intervals in the analysis of heart rate variability. Front Physiol 3: 148, 2012.
24. Pöyhönen M, Syväoja S, Hartikainen J, Ruokonen E, Takala J. The effect of carbon dioxide, respiratory rate and tidal volume on human heart rate variability. Acta Anaesthesiol Scand 48: 93–101, 2004.
25. Quintana DS, Heathers JA, Kemp AH. On the validity of using the Polar RS800 heart rate monitor for heart rate variability research. Eur J Appl Physiol 112: 4179–4180, 2012.
26. Salo MA, Huikuri HV, Seppanen T. Ectopic beats in heart rate variability analysis: Effects of editing on time and frequency domain measures. Ann Noninvasive Electrocardiol 6: 5–17, 2001.
27. Sami S, Seppänen M, Kuusela A. Artefact Correction for Heart Beat Interval Data. Presented at 1st Advanced Methods for Processing Bioelectrical Signals Meeting: University of Jyväskylä; October 1, 2004; Jyväskylä, Finland.
28. Tarvainen MP, Niskanen J-P, Lipponen JA, Ranta-Aho PO, Karjalainen PA. Kubios HRV–heart rate variability analysis software. Comput Methods Programs Biomed 113: 210–220, 2014.
29. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Standards of measurement, physiological interpretation, and clinical use. Circulation 93: 1043–1065, 1996.
30. Thomas JR, Nelson JK, Silverman SJ. Research Methods in Physical Activity. Champaign, IL: Human Kinetics, 2010.
31. Tulppo MP, Hughson RL, Mäkikallio TH, Airaksinen KJ, Seppänen T, Huikuri HV. Effects of exercise and passive head-up tilt on fractal and complexity properties of heart rate dynamics. Am J Physiol 280: H1081–H1087, 2001.
32. Vanderlei L, Silva R, Pastre C, Azevedo FM, Godoy M. Comparison of the Polar S810i monitor and the ECG for the analysis of heart rate variability in the time and frequency domains. Braz J Med Biol Res 41: 854–859, 2008.
33. Wallén MB, Hasson D, Theorell T, Canlon B, Osika W. Possibilities and limitations of the polar RS800 in measuring heart rate variability at rest. Eur J Appl Physiol 112: 1153–1165, 2012.
34. Weippert M, Behrens M, Rieger A, Behrens K. Sample entropy and traditional measures of heart rate dynamics reveal different modes of cardiovascular control during low intensity exercise. Entropy 16: 5698–5711, 2014.
35. Weippert M, Kumar M, Kreuzfeld S, Arndt D, Rieger A, Stoll R. Comparison of three mobile devices for measuring R–R intervals and heart rate variability: Polar S810i, Suunto t6 and an ambulatory ECG system. Eur J Appl Physiol 109: 779–786, 2010.
36. Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res 19: 231–240, 2005.