Temporal cross-correlation between Polar ® heart rate monitor interface board and ECG to measure RR interval at rest

Abstract

BACKGROUND:

A Polar heart rate monitor is a device that measures RR intervals, but has not been correlated to accurately measure the series of RR interval signals between the ECG and the Polar ${}^{\text{®}}$ Heart Rate Monitor Interface (HRMI) Board at rest.

OBJECTIVE:

The aim of the study was to evaluate the temporal cross-correlations between the Polar ${}^{\text{®}}$ HRMI Board and an ECG to measure the series of RR intervals at rest.

METHODS:

The sample consisted of eighteen healthy male subjects and they were instructed to lie in the supine position at rest while breathing normally and a time window of the last 2 min was recorded to analyse RR intervals were obtained for each subject with a Polar ${}^{\text{®}}$ HRMI Board and an ECG. Cross-correlation analysis of RR interval signals between methods and reliability was expressed by Bland and Altman analysis.

RESULTS:

The cross-correlation was excellent, resulting in a mean of 0.98 $\pm$ 0.01 and no lag or delay between the signals. The bias was 0.03 $\pm$ 0.08 s or 8.0% for MeanRRi from Polar ${}^{\text{®}}$ HRMI Board and ECG, no significant difference.

CONCLUSIONS:

The Polar ${}^{\text{®}}$ HRMI Board is acceptable for assessment of serial RR intervals. The results support the reliability of the Mean RR interval compared to a resting ECG.

1. Introduction

Heart rate variability (HRV) is a non-invasive tool used to assess the autonomic control of heart rate fluctuations [1]. In the sports field, HRV has been found to be a valuable measure in a variety of sports settings with the measurement of many factors including overtraining [2], heart dynamics and efforts recovery [3], exercise prescription [4], identification of anaerobic thresholds [5] and determination of an athlete’s fitness level [6].

The use of electrocardiogram (ECG) has been used to study HRV analysis, but the high cost, difficulty of application in situations outside the laboratory environment, such as in physical training conditions [7, 8] make this a rather impractical tool for use outside the laboratory. As an alternative, the Polar ${}^{\text{®}}$ Heart Rate Monitor Interface Board has overcome these difficulties, but it is basically a heart rate monitor that records the beat-to-beat RR intervals (RRi) of the ECG signal [9, 10]. The device is able to generate an output of a of 1 ms pulse, indicating a cardiac contraction [11].

Previous studies have investigated the correlation between commercial ECG systems at rest [12, 13, 14, 15, 16, 17] and during exercise [18, 19, 20, 21, 22] based on HRV. In addition, none of these studies investigated the use of the cross-correlation to accurately measure the series of RR intervals between the ECG and the Polar ${}^{\text{®}}$ HRMI Board at rest, which is still an open area for investigation.

The cross-correlation function [23] estimates the correlation between the series allocated at time ( $t$ ) and the second series at time ( $t+k$ ), where $k$ is the delay or number of lags in the series according to Eq. (1).

$R_{xy}(k)={\displaystyle \sum _{k=-\mathrm{\infty }}^{\mathrm{\infty }}}X(t)Y(t+k)$(1)

Cross-correlation is a way of measuring the degree of similarity between a time series and a lagged version of another time series [24] or the interdependence between the signals [25]. It can also be used to determine phase differences, extract hidden signals, and reveal frequency content [25]. Thus, the aim of the study was to temporal cross-correlation between Polar ${}^{\text{®}}$ Heart Rate Monitor Interface Board and ECG to measure RR interval at rest.

2. Methods and materials

2.1 Subjects

The sample consisted of eighteen adult male athletes 20 to 25 yrs of old, who volunteered for this study. All subjects were physical education students, non-smokers, with no history of cardiopulmonary disease and none was taking any medication, randomly selected during an anthropometric evaluation in the Physical Education course at the Federal University of Amapá (UNIFAP) between May and December 2022 in the period from 08:00 to 10:00 hours.

The study protocol was approved by the local Human Research Ethics Committee of the Federal University of Amapá (CAAE: 50150121.1.0000.0003, n ${}^{\circ }$ 5.121.013), and an informed written consent was obtained from all subjects. This study was conducted in accordance with the instructions of the Helsinki Declaration of 2008 and in accordance with Resolution 510/2016 of the National Health Council.

2.2 Anthropometric measurements

Potential volunteers were given a verbal explanation of the testing procedures and the time commitment required to participate in this study during an orientation session. Data collection and the anthropometric variables were performed by the same and experienced evaluator throughout the study. For both weight and height measurements, participants were kept barefoot, wearing light clothes and not carrying any object. The height was measured in centimeters and body weight in kilograms using certified and calibrated mechanical scales (Filizola, Brazil).

2.3 Data collection

The tests were conducted in a quiet room with temperature maintained at 22 ${}^{\circ }$ C. All subjects were instructed to avoid strenuous activity in the 24-h prior to each testing session and to avoid alcohol, caffeine, and the consumption of large meals for at least 3-h prior to testing. At the first visit to the laboratory, all subjects were instructed to lie in the supine position for 5-min at rest while breathing normally and a time window of the last 2-min was recorded for analysis RR intervals were obtained for each subject using a Polar ${}^{\text{®}}$ HRMI Board and ECG.

The skin was prepared by shaving the hair, abrasion with sandpaper and alcohol cleaning. The ECG signal was measured by an electrocardiograph MCL07 (Ecafix, Brazil) in the modified lead CM5 to improve signal quality and R-wave amplitudes and surface electrodes of silver-silver chloride (Ag/AgCl) disposable Kendall Meditrace 200 (The Ludlow, Chicopee, USA). Additionally, cardiac pulses were also collected through Polar ${}^{\text{®}}$ HRMI Board (Sparkfun, USA). Pulses were transmitted to the HRMI by a Polar ${}^{\text{®}}$ (T31, Kenpele, Finland) belt placed around the chest of each subject. All signals were acquired and simultaneously digitized using a data acquisition Board model NIUSB-6008 (National Instruments, USA), with a 12-bit resolution analog-to-digital converter and a dynamic range $\pm$ 5 V, providing a quantization error of 1.22  $\mu$ V with the gain 1000, at a sampling rate of 1000 Hz and stored on a microcomputer for off-line processing. All procedures for the acquisition and control data were performed in the software DAS [26], built in LabVIEW version 5.0 (National Instruments, USA) in a microcomputer with Windows 7 operating system (Microsoft, USA).

2.4 Signal processing

All signal processing was analyzed with programs written in Matlab version 2020.b (The MathWorks, USA). To measure the RRi of the ECG, the signal was preprocessed by a 2th order Butterworth band-pass filter with cutoff frequencies of 5 and 20 Hz, applied in both the direct and reverse directions of the signals to avoid phase distortions. The times of the peaks of R-waves were detected by the zero crossing of the derivative of the filtered signal. Thus, the algorithm selected all signal peaks above a threshold of 0.5 V and removed the peaks whose refractory period was less than 250 ms. In addition, the algorithm allowed manual editing with visual inspection, in order to exclude ectopic heartbeats and exported as RR intervals in seconds.

A pulse train of peaks from the Polar ${}^{\text{®}}$ HRMI Board was then generated to measure the RRi from the Polar ${}^{\text{®}}$ HRMI Board. First, the algorithm selected all signal peaks above an established threshold, derived the signal and divided the result by the sampling frequency to find the RRi, which were automatically corrected using a moving average filter. The detected RRi were manually assessed to ensure that they had been correctly detected. Ectopic beats were noted and corrected by the Outliers RR module based on the first and third quartile intervals. All records of sample data had less than 2% error and were then saved to .txt files and exported as RR intervals in seconds.

2.5 Time series analysis

The cross-correlation analysis [24] between the RR interval signals from the ECG and the Polar ${}^{\text{®}}$ HRMI Board at rest in the time domain according to Eq. (2):

$C_{\text{𝑖𝑅𝑅𝐸𝐶𝐺},\text{𝑖𝑅𝑅𝑃𝑜𝑙𝑎𝑟}}(t)=$

(2) ${\displaystyle \sum _{k=-\mathrm{\infty }}^{\mathrm{\infty }}}\text{𝑖𝑅𝑅𝐸𝐶𝐺}(t)\text{𝑖𝑅𝑅𝑃𝑜𝑙𝑎𝑟}(t+k)$

where $k$ is the time lag.

For each participant, the mean of the RR intervals (MeanRRi) from the Polar ${}^{\text{®}}$ HRMI Board and the ECG at rest was processed and finally the reliability between the HRV parameter in the time domain was studied.

2.6 Statistical analysis

Descriptive data were expressed as mean $\pm$ standard deviation (SD). The Lilliefors test confirmed the normality of the sample distribution. Post hoc power analysis (1 $-$ beta error level) was used to determine the effect size [27] using G*Power software version 3.1.9.2 (University of Kiel, Germany). The MeanRRi data obtained from the Polar ${}^{\text{®}}$ HRMI Board and ECG were estimated for each subject as absolute and relative numbers. Subsequently, the Student t-test for dependent samples was used to test the differences between the MeanRRi absolute data obtained from the Polar ${}^{\text{®}}$ HRMI Board and an ECG with subjects at rest. To analyze the agreement of the MeanRRi data for each subject at rest obtained from the Polar ${}^{\text{®}}$ Heart Rate Monitor Interface Board compared to an ECG, we followed the method Bland and Altman method [28] calculating the limits of agreement (LOA) at 95 % based on the mean difference, expressed as $\pm$ 2 SD. All procedures assumed $p⩽$ 0.05 for statistical significance. All data were processed in Matlab version 2020.b (The MathWorks, USA).

3. Results

Table 1

Anthropometric and physical characteristics and mean RRi obtained from ECG and Polar ${}^{\text{®}}$ HRMI Board of participants at rest

Variables	Mean $\pm$ SD	CI95%	$P$ -value
Age (years)	23.8 $\pm$ 2.9	22.0–25.6	0.15
Height (cm)	176.1 $\pm$ 6.9	172.0–180.3	0.14
Body mass (kg)	80.6 $\pm$ 11.5	73.7–87.5	0.79
MeanRRi ECG	0.88 $\pm$ 0.12	0.82–0.94	0.26
MeanRRi Polar	0.91 $\pm$ 0.11	0.85–0.96	0.08

Values are mean $\pm$ standard deviation (SD), CI95% is confidence intervals 95% and $P$ -value of Lilliefors test.

Figure 1.

Cross-correlation RR interval signals between Polar ${}^{\text{®}}$ HRMI Board and ECG in one subject. Lags is $N$ $-$ 1: $N$ $+$ 1 and $N$ is the total sample.

Anthropometric, physical characteristics and MeanRRi from the Polar ${}^{\text{®}}$ HRMI Board and ECG during rest of the participants are presented in Table 1, for all the variables was a Gaussian distribution ( $p⩾$ 0.05). The sample size effect was considered large with an actual power of 0.99.

The MeanRRi data obtained from the Polar ${}^{\text{®}}$ HRMI Board and ECG varied for each subject in both absolute and relative terms, resulting in a mean difference of 0.03 $\pm$ 0.08 s or 8.0%, which was not significantly different ( $p=$ 0.46).

Cross-correlation between Polar ${}^{\text{®}}$ HRMI Board and ECG RR interval signals was excellent, resulting in means of 0.98 $\pm$ 0.01 and no lag or delay between the signals, as shown in Fig. 1.

The Bland and Altman [28] plot analyzed the agreement resting MeanRRi data between the Polar ${}^{\text{®}}$ HRMI Board and the ECG. Regardless of the reliable 95% LOA, only 5.5% of the coordinates of the MeanRRi variable on the HRV parameters were expressed outside the LOA, homoscedasticity of the data as shown in Fig. 2.

Figure 2.

Bland and Altman [25] plot of the MeanRRi HRV parameter between Polar ${}^{\text{®}}$ HRMI Board and ECG, where the middle line represents the mean difference between the methods and the outer lines represent the 95% limit of agreement (LOA), calculated as $\pm$ 2 standard deviations.

Figure 3.

RR interval signals between Polar ${}^{\text{®}}$ HRMI Board (gray) and ECG (black) in one subject during the last 2 minutes at rest.

4. Discussion

The purpose of this study was to determine the temporal cross-correlation between the Polar ${}^{\text{®}}$ Heart Rate Monitor Interface Board and ECG to measure RR intervals at rest. It was hypothesized that the correlation is essential to accurately measure the series of RR intervals between the ECG and the Polar ${}^{\text{®}}$ HRMI Board.

Previous studies [12, 13, 14, 15, 16, 17] support the validity of the Polar ${}^{\text{®}}$ HRMI Board for measuring RR intervals and make the subsequent resting HRV analysis. However, in contrast to the present study, none of these studies investigated the temporal cross-correlations between the Polar ${}^{\text{®}}$ HRMI Board and the ECG to measure serial RR intervals.

The time domain analysis based on the cross-correlation between these RR interval signals shows the presence of correlation between the rhythms, as it resulted in excellent cross-correlation, with an average of 0.98 $\pm$ 0.01 peak values and, no lag or delay between the signals. The results suggest that the Polar ${}^{\text{®}}$ HRMI Board can produce RR interval recordings is consistent with an ECG and that the MeanRRi HRV parameter derived from these recordings is comparable in adult male athletes at rest.

The RR interval is the time elapsed between two successive R-waves of the QRS signal on the ECG is a function of the intrinsic properties of the sinus node as well as autonomic influences [29]. For this reason, MeanRRi is the average RR interval duration in a measurement. Autonomic cardiovascular regulation can be studied by quantifying MeanRRi as assessed under steady state conditions of health [30, 31] or disease [32, 33] when compared to other HRV parameters, which justifies the use of the MeanRRi HRV parameter in the study. However, no statistically significant difference was found when testing the differences in the MeanRRi HRV parameter obtained during at rest between the Polar ${}^{\text{®}}$ HRMI Board and an ECG recorder.

Bland and Altman [34] have investigated about the use of correlation is misleading, an alternative approach, based on graphical techniques and simple calculations is described along with the relationship between this analysis and the assessment of repeatability. Therefore, the Bland and Altman [28] plot illustrates the coordinates of MeanRRi of the HRV parameter, which showed a low variability and few coordinates outside the limits of agreement, expressed as 95% LOA, indicating homoscedasticity or homogeneity of variances.

Figure 3 illustrates the RR interval signals between the Polar ${}^{\text{®}}$ HRMI Board and ECG in one subject during the last 2 minutes at rest, as observed greater amplitude variation RR interval data of the Polar ${}^{\text{®}}$ HRMI Board when compared to an ECG recorder, the mean RR differences between devices in this subject was about 0.1 s, but the difference in means was of 0.03 $\pm$ 0.08 s or a mean relative error of 8.0%. Since the Polar ${}^{\text{®}}$ HRMI Board system did not satisfactorily detect errors or provide valid HRV values for certain groups, it is concluded that traditional ECGs should be used whenever possible for both acquisition and processing of HRV data [35].

The present study was limited to the study of the MeanRRi HRV parameter alone in the time domain. Future research should address the limitations of the present study by employing a sample of different age and both sexes, analysis of HRV parameters in the frequency domain with respiratory rate control [36] and control group by investigating the associations between these parameters HRV values and cardiovascular risk factors.

5. Conclusions

The Polar ${}^{\text{®}}$ HRMI Board is considered acceptable to assess the series RR intervals and the findings support the reliability to produce MeanRRi recordings consistent when compared with an ECG at rest.