BIOMEDICAL ENGINEERING Recent Developments Homer Nazeran, Michael Goldman, Richard Schoephoerster, Editors © 2008 Medical and Engineering Publishers, Inc., ISBN 978-1-930636-06-4 HEART RATE VARIABILITY SIGNAL PARAMETERS QUANTIFY SKIN COOLING EFFECT OF ENERGY PATCHES DURING REST AND EXERCISE IN YOUNG HEALTHY INDIVIDUALS Homer Nazeran PhD, CPEng (Biomed.) Electrical and Computer Engineering, The University of Texas at El Paso, El Paso Texas 79968 ABSTRACT Heart rate variability (HRV) signal analysis provides a non-invasive and sensitive marker of autonomic nervous system (ANS) activity. Spectral parameters of HRV signal are used to quantify the balance between sympathetic and parasympathetic (sympathovagal) influences under various physiologic conditions. ECG signals were acquired, filtered and further processed to derive the HRV signal. The low frequency (LF), high frequency (HF), and their ratio LF/HF were calculated to assess the parasympathetic dominance or the skin cooling effect of a set of non-transdermal Energy Patches on young healthy individuals during Rest and immediately after mild Exercise. HRV data acquired from 20 young healthy volunteers (10 males and 10 females, 19-25 years of age), in a doubleblind placebo-controlled study, were used to evaluate the skin cooling effect of these wearable devices on the ANS during rest and immediately after mild exercise while wearing active (A) and placebo (P) patches. Data from condition (A) and condition (P) were compared using statistical analysis (one-sample inference). The LF/HF decreased significantly both during rest and immediately after mild exercise in condition (A) compared to condition (P) with derived from spectral analysis of HRV signals could be used to quantify the parasympathetic dominance or the localized skin cooling effect of non-transdermal energy patches during rest and immediately after mild exercise in young healthy individuals. Keywords: Heart rate variability signal analysis, Localized skin cooling effects, Non-transdermal energy patches, Autonomic nervous activity, Wearable devices INTRODUCTION The autonomic nervous system (ANS) is structurally and functionally subdivided into two antagonistic divisions: sympathetic and parasympathetic. The sympathetic division responds to fight-or-flight situations, increasing heart rate (HR) and blood pressure (BP), among others, under stressful conditions. The parasympathetic division is responsible for relaxation and energy conservation (i.e., decreasing HR and BP). The heart as well as other organs, receives opposing influences from these two arms of the ANS. This dual innervation underpins a fine and continuous regulatory system and simply means an increase in the activity of one division results in a smooth and reciprocal decrease in the activity of the other. The dynamic interplay between the two divisions increases or decreases the HR depending on the predominance of one over the other. Heart Rate Variability (HRV) signal refers to beat-to-beat variation of heart rate and represents the cyclical changes in HR. As HR is modulated by both parasympathetic and sympathetic inputs, HRV can be utilized as an indirect and non-invasive marker of autonomic regulation and control under different physiological conditions [1]. High HRV reflects an ANS that is adaptable and dynamically responsive to change whereas reduced HRV is indicative of an abnormal or restricted ability of the ANS in maintaining homeostasis [2, 3]. Pharmacological studies and spectral analysis of the HRV signal have revealed two clear peaks in its power spectrum: a high frequency (HF) and a low frequency (LF) component. The HF peak which is typically centered around 0.25 Hz (0.15 – 0.4 Hz) arises from the coupling of the respiratory and cardiac cycle and is termed Respiratory Sinus Arrhythmia (RSA). Vagal 1 blockade abolishes this HF peak suggesting that it is parasympathetically mediated [3]. The LF peak, which is typically centered around 0.1 Hz (0.05 – 0.15), arises from BP changes (Mayer-waves) and can be modified by both vagal and sympathetic blockade and therefore reflects both sympathetic and parasympathetic influences [3]. A better measure of sympathetic modulation is the LF/HF which is indicative of the balance between sympathetic and parasympathetic activity (sympatho-vagal balance) with a low frequency bias suggesting sympathetic dominance [1, 3]. A number of research studies in recent years have confirmed that higher variability in HR is indicative of a healthy ANS, while reduced variations in the HR is a sign of diseased or pathological state. With an increased number of the population living longer, novel technologies are emerging to decrease the degenerating effects of the aging process. The average citizen desires to continue to be active and robust as he/she grows old. This attitude towards aging has led to the development of a number of novel wearable devices to promote increased energy, vitality, pain relief, and mental focus, with the objective to prolong and improve the quality of life. One such device is a set of non-transdermal skin energy patches comprised of a white patch and a tan patch that is claimed to improve energy, stamina and performance [4]. The very small (“nanoscale”) optically active chiral molecules in these non-transdermal energy patches passively absorb significant amounts of heat (“infrared - wide band”) energy emanating from the human body in the area covered by the patches and then reemit only a small portion (“narrow band”) of this energy back into the body through selective acupuncture points which allow efficient reflection of “narrow band” heat energy into the body. As these patches actively trap a good portion of the infrared energy emitted by the body and reflect back only a small portion of this energy, they first create a localized thermal cooling effect on the body. The localized thermal cooling effect of the patches next influences the autonomic nervous system and modifies the sympatho-vagal balance in favor of parasympathetic influence (skin cooling effect) resulting in regional cooling of warm areas. This thermal cooling effect on the body has been measured and mapped using infrared imaging [5]. This is shown in Figure 1. This effect is present during rest and is enhanced during exercise when more heat is generated by the body. The improvement in exercise performance due to skin pre-cooling effect is very well established in the sports medicine and physiological literature [6 -14]. FIGURE 1. Infrared images showing the skin cooling effect of non-transdermal patches both local and distal to their application site. The upper and lower left images show the body heat distribution patterns of the upper thoracic paraspinal region upon application of the patches (upper) and 5 minutes after their application (lower). The images on the upper and lower right show the body heat distribution patterns of the abdominal region upon application of the patches (upper) and the 5 minutes after their application (lower) [5]. (Courtesy of Dean Clark, DC) 2 To investigate the effect of these wearable devices on the ANS, a double-blind placebo-controlled study was carried out. HRV data acquired from 20 young healthy volunteers (10 males and 10 females, age range: 19-25 years), were used during rest and immediately after mild exercise while wearing active (A) and placebo (P) patches. Data from condition (A) and condition (P) were then compared. The average LF/HF of the HRV signal decreased significantly both during rest and immediately after mild exercise in condition (A) compared to condition (P) with an <0.01. This study shows that the LF/HF derived from spectral analysis of the HRV signal quantifies the parasympathetic dominance or the skin cooling effect of trans-dermal energy patches during rest and immediately after mild exercise in young healthy individuals. MATERIALS AND METHODS ECG data were recorded from 20 young healthy volunteers using an IRB approved protocol. Two sets of patches (Placebo and Energy) were provided to the participants by a researcher totally blind to the two sets of patches marked: Set-1 and Set-2. Each set comprised of a white patch and a tan patch. Instructions were provided to the participants to apply the patches bilaterally over one of four possible anatomical landmarks with the white patch worn on the right side of the body and the tan patch worn on the left side of the body (see Figure 3). A set of Placebo (P) patches: patches with inactive natural nutritional substances or preparation (i.e., normal saline solution), were used as a control to determine the effectiveness of the Active (A) or Energy patches during Rest and immediately after mild Exercise. A set of Energy patches: patches that allow natural biomolecules encased within them to interact with the human body heat were used as active (A) patches. ECG data acquired from the volunteers were used to derive the HRV signals. The participants were instructed to drink plenty of water and feel fully hydrated before data collection. Data were recorded under six different conditions: 1) at rest; 2) after 5 min exercise (walking up and down stairs); 3) at Rest wearing the Placebo patches; 4) after 5 min mild Exercise wearing the Placebo patches; 5) at Rest wearing the Energy patches; 6) after 5 min mild Exercise wearing the Energy patches. FIGURE 2. The white patch and tan Energy patches FIGURE 3. The four possible bilateral anatomical locations for patch placement To perform short-term HRV signal analysis, ECG signals from each volunteer were recorded for 5 minutes [1] during each condition using either a custom-made PDA-based ECG data acquisition system or a Student Biopac Lab system (www.biopac.com). Three disposable Ag/AgCl ECG electrodes were used to sense the ECG signals. ECG data were recorded under 6 different conditions using a sampling rate of 360 Hz. Accurate determination of the QRS complex and more specifically, reliable detection of the R-wave peak plays a central role in computerbased ECG signal analysis, Holter data analysis and robust HRV signal analysis. Fluctuations in the ECG signal baseline drift, motion artifact due to electrode movement and electromyographic (EMG) interference due to muscular activity frequently contaminate ECG signals. In addition, morphological variations in ECG waveform and the high degree of heterogeneity in QRS complexes make it difficult to distinguish them from tall-peaked P and T waves. An Enhanced Hilbert Transform algorithm with automatic R-peak correction capability was used to detect the QRS complexes. Occurrence of R-peaks in the QRS complexes were manually verified for error-free detection. 3 In this algorithm, band-pass filtered ECG signals were first differentiated to reduce baseline drift and motion artifacts further. With implementation of this step, R-peaks were set to zero. Hilbert Transform (HT) of the differentiated signal was then calculated and its conjugate was determined. This step made the zero crossings in the differentiated signal prominent peaks in the Hilbert transformed conjugate of the differentiated signal. Using the normalized RMS value of the Hilbert transformed ECG signal a threshold value was determined. The time instants for which the signal amplitude was greater than the threshold value were stored in an array. Peak detection was performed on the original signal. The peak values and the times at which they occurred were stored in separate arrays. The RR intervals were then calculated. If an R-R interval was greater than 130% of the previous R-R interval, then the HRV signal was considered to have missed an R-peak from the ECG signal. Hence a correction for the missing beat was made based upon an updated moving average of the previous R-R intervals. The amplitude of the missing peak was estimated as the average of the two previous R-peaks adjacent to it [15]. As the HRV signals were not equally sampled, an interpolation was performed to achieve uniform sampling before carrying out spectral analysis. The cubic spline interpolation technique was deployed for this purpose to enable re-sampling at desirable frequencies. Cubic splines are not band-limited, therefore to avoid aliasing the interpolated HRV signals were re-sampled at 2 Hz. A lowpass Butterworth filter with a cutoff frequency of 1 Hz was then used to eliminate the frequencies above 1 Hz. The re-sampled HRV signals were finally decimated by a factor of 2 [15]. An example of the missing-peak detection capability of the algorithm deployed in this study is shown in Figure 4. FIGURE 4. An example of the missing peak detection capability of the Enhanced Hilbert transform algorithm [15] Power spectral analysis of HRV signals provides the basic information on how power within the signal is distributed as a function of frequency. Two general methods of HRV spectral analysis are in popular use: nonparametric and parametric. The parametric method was deployed in this study for the following reasons: 1. it provides smoother spectral components which can be distinguished independently, 2. it facilitates easy post–processing of the spectrum, 3. it facilitates the identification of the central frequency of each component, and 4. it provides an accurate estimation of the power spectrum with small number of sample points (short record length) when the statistical characteristics of the signal remains stable (signal remains stationary). Variations in the HR, occurring at the spectral frequency band of 0.15- 0.4 Hz, known as high frequency (HF) band, are indicative of parasympathetic (vagal) activity. Variations in the spectral band 0.05-0.15 Hz, known as low frequency (LF) band, are linked to the sympathetic modulation, but this component includes some parasympathetic influence (sympatho-vagal influences). The spectral band 0.003-0.05 Hz, known as very low frequency (VLF), is possibly related to the long-term regulatory mechanisms. It is well established that the level of physical activity is clearly reflected in the HRV power spectrum. For example, when a healthy subject stands up, there is an increase of HRV in the LF spectral band, which is considered to be an estimate of the sympathetic influence on the heart. Consequently, the LF/HF is a quantitative indication of sympathovagal balance or sympathetic modulations [1]. 4 The spectral components, namely the VLF, LF, HF and LF/HF of the HRV signals, as well as a number of their time-domain, frequency-domain and nonlinear dynamics measures were calculated from 5-minute ECG recordings under 6 different conditions while wearing P and A patches. Power spectrum of HRV signals were performed using the autoregressive (AR) spectral estimation method. A model order of p = 20 was used for this purpose [15]. A custom-developed integrated software package with a user-friendly graphical-user-interface (GUI) shown in Figure 5 was used to perform comprehensive HRV signal analysis [17]. The power spectral components of the HRV signal were determined in normalized units (n.u.) which represent the relative values of each power component in proportion to the total power minus the VLF component. The representation of LF and HF in n.u. emphasizes the controlled and balanced behavior of the two branches of the ANS [1]. It has been shown that LF and HF can increase under different physical and physiological conditions. An increase in normalized LF is observed during 90o tilt, standing, mental stress and moderate exercise in healthy subjects [1]. Conversely, an increase in normalized HF is induced by controlled respiration, cold stimulation of the face and rotational stimuli (See references in [1]). FIGURE 5. User-friendly GUI for a custom-made integrated software environment to perform comprehensive HRV signal analysis [17]. The values shown are from data used in a children’s sleep study. RESULTS Table 1 shows the normalized spectral parameters calculated from the power spectrum of 5-minute HRV signals acquired from the ECGs of a healthy young male volunteer under 6 different conditions. Table 2 shows the normalized spectral parameters calculated from the power spectrum of 5-minute HRV signals acquired from the ECGs of a healthy young female volunteer under 6 different conditions. The overall decrease in the mean LF/HF n.u. in the subjects (n = 20) when wearing the Energy patches (compared to when wearing the Placebo patches) during Rest was 42% with a standard deviation of 53% of the mean. The overall decrease in the mean LF/HF n.u. in the subjects (n = 20) when wearing the Energy patches (compared to when wearing the Placebo patches) immediately after 5 minutes of mild Exercise was 58% with a standard deviation of 49% of the mean. With these values a statistical power of at least 85% at a significance level or p < 0.01 was achieved. This level of significance reflects a very significant effect. As a value of or p = 0.05 is considered statistically significant, the results in these subjects reflect a statistically significant effect with a power of 90%. 5 TABLE 1. Typical normalized power spectral parameters derived from 5-minute HRV signals acquired from ECGs of a healthy male volunteer under 6 different conditions [16] Condition LF n.u. HF n.u. LF/HF n.u. Rest with no Patches Rest with Placebo Patches Rest with Energy Patches 97.70 97.62 96.95 2.28 2.31 3.01 42.75 42.23 32.29 Exercise with no Patches Exercise with Placebo Patches Exercise with Energy Patches 97.42 97.97 94.72 2.56 2.03 5.23 38.03 48.36 18.10 TABLE 2. Typical normalized power spectral parameters derived from 5-minute HRV signals acquired from ECGs of a healthy female volunteer under 6 different conditions [16]. Condition LF n.u. HF n.u. LF/HF n.u. Rest with no Patches Rest with Placebo Patches Rest with Energy Patches 99.15 97.33 94.20 0.83 2.58 5.37 119.25 37.61 17.54 Exercise with no Patches Exercise with Placebo Patches Exercise with Energy Patches 90.58 96.40 95.32 9.41 3.54 4.62 9.62 27.23 20.63 DISCUSSION AND CONCLUSION From the spectral parameters of typical datasets and the statistical analysis results for 20 subjects, the following observations could be made: 1. There was a noticeable difference between spectral parameters when the subjects wore the Placebo patches compared with when the subjects wore no patches. Such differences in spectral parameters are indicative of the Placebo Effect. This is an indication of how the subjects responded to the feeling of wearing a patch. 2. There was a decrease in the normalized LF when the subjects wore the Energy patches compared with the condition when the subjects wore the Placebo patches during both Rest and immediately after 5 minutes of mild Exercise. 3. There was an increase in the normalized HF when the subjects wore the Energy patches compared with the condition when the subjects wore the Placebo patches during both Rest and immediately after 5 minutes of mild Exercise. 4. There was a statistically significant decrease in the normalized LF/HF when the subjects wore the Energy patches compared with the condition when the subjects wore the Placebo patches during both Rest and immediately after 5 minutes of mild Exercise. 5. On average, female subjects were more responsive to Energy patches compared to male subjects during Rest. While, on average male subjects were more responsive to Energy patches compared to females after mild Exercise. These differences were not statistically significant. 6. On average, there was a higher reduction in normalized LF/HF after 5 minutes of mild Exercise compared to the reduction in normalized LF/HF during Rest. Based on these observations it could be concluded that both during Rest and immediately after 5 minutes of mild Exercise, the Energy patches elicited an enhanced parasympathetic response (due to a localized skin cooling effect) which could be quantified by a reduction in normalized LF/HF. A further reduction of normalized LF/HF immediately after 5 minutes of mild Exercise (as a consequence of more body heat production) compared to Rest may be indicative of the higher activation level of the Energy patches in response to enhanced physical activity resulting in an increased localized skin cooling effect during Exercise. The statistical results revealed that the Energy patches showed a very significant effect (p < 0.01) compared to Placebo patches in reducing the normalized LF/HF during Rest and even further after 5 minutes of mild Exercise with a statistical power of at least 85% in this sub-population. 6 REFERENCES 1. Task Force of the European Society of Cardiology and The North American Society of Pacing and Electro physiology. Heart rate variability. Standards of measurement, physiological interpretation, and clinical use. Eur Heart J 1996; 17: 354-381 2. McMillan, DE. Interpreting heart rate variability sleep/wake patterns in cardiac patients. J Cardiovas Nurs 2002; 17: 69-81 3. Pumprla, J, K Howorka, D Groves, M Chester, J Nolan. Functional assessment of heart rate variability: physiological basis and practical applications. Intl J Cardiol 2002; 84:1-14 4. http://www.lifewave.com/ 5. http://www.lifewave.com/pdf/Research/Research003-SummaryIceWaveClinical.pdf 6. Grahn, DA, VH Cao, et al. Heat extraction through the palm of one hand improves aerobic exercise endurance in hot environment. J Appl Physiol 2005; 99:972-978 7. Arngrismsson SA, DS Pettit, et al,. Cooling vest worn during active warm-up improves 5-km run performance in heat. J Appl Physiol 2004; 96: 1967-1874 8. Duffield, R, B Dawson, et al,. Effect of wearing an ice cooling jacket on repeat sprint performance in warn/humid conditions. Br J Sports Med 2003; 37: 164-169 9. White, AT, SL Davis, et al,. Metabolic, thermoregulatory, and perceptual responses during exercise after lower vs. whole body precooling. J Appl Physiol 2003; 94: 1039-1044 10. Marino, FE. Methods, advantages, and limitations of body cooling for exercise performance. Br J Sports Med 2002; 36:89-94 11. White, AT, TE Wilson, et al. Effect of precooling on physical performance in multiple sclerosis. Multiple Sclerosis2000; 6: 176-180 12. Galloway SDR, RJ Maughan. Effects of ambient temperature on the capacity to perform prolonged cycle exercise in man. Med Sci Sports Exer 1997; 29(9): 1240-1249 13. Bruck, K, H Olschewiski, et al,. Body temperature related factors diminishing the drive to exercise. Can J Physiol Pharmacol 1987; 65(6): 1274-1280 14. Hessemer, V, D Langusch, et al. Effects of slightly lowered body temperatures on endurance performance in humans. J Appl Physiol 1984; 57(6): 1731-1737 15. Chatlapalli, S, H Nazeran, et al. Accurate Derivation of Heart Rate Variability Signal for Detection of Sleep Disordered Breathing in Children. 26th Annual International Conference of the IEEE EMBS. 2004; 538-541 16. Nazeran, H, M Chatlapalli, et al. Effect of Novel Nanoscale Energy Patches on Spectral and Nonlinear Dynamic Features of Heart Rate Variability Signals in Healthy Individuals during Rest and Exercise. 27th Annual International Conference of the IEEE EMBS, 2005 17. Nazeran H, R Krishnam, et al. Nonlinear Dynamics Analysis of Heart Rate Variability Signals to Detect Sleep Disordered Breathing in Children. 28th Annual International Conference of the IEEE EMBS. 2006; 3873-3878 7