1 Heart rate variability in the acceleration photoplethysmogram at rest and after exercise—a preliminary study Mohamed Elgendi 1,* , Socrates Dokos 2 , Derek Abbott 3 1 Department of Computing Science, University of Alberta, Edmonton, Alberta, Canada 2 Graduate School of Biomedical Engineering, University of New South Wales, Sydney, New South Wales, Australia 3 School of Electrical and Electronic Engineering, University of Adelaide, Adelaide, South Australia, Australia * E-mail: [email protected]Abstract There are a limited number of studies on heart rate variability (HRV) dynamics immediately after exer- cise. The electrocardiogram (ECG) may be used to measure HRV, however acquiring ECG signals from subjects undergoing exercise is not convenient. Many researchers have demonstrated that photoplethys- mogram (PPG) signals offer an alternative method to measure HRV when ECG and PPG signals are simultaneously collected. However, we investigate a different approach to potentially show that the PPG signals can measure HRV without collecting ECG signals. Moreover, we explore the extraction of the most suitable HRV-parameters from short PPG signal recordings. Our preliminary results now motivate further studies that cross check HRV parameters extracted from both ECG and PPG. In this study, PPG signals from an existing database were used to determine a range of HRV indices, including the standard deviation of heart beat interval (SDNN) and the root-mean square of the difference of successive heart beats (rMSSD). Results from this study indicate that the use of the a–a interval, derived from the ac- celeration of PPG signals, show very promising results in determining the HRV statistical indices SDNN and rMSSD over 20-second PPG recordings. Moreover, post-exercise SDNN and rMSSD indices show negative correlation with age.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Heart rate variability in the acceleration photoplethysmogram
13. Task Force of the European Society of Cardiology the North American Society of Pacing and
Electrophysiology (1996) Heart rate variability: Standards of measurement, physiological interpre-
tation, and clinical use. Circulation 93: 1043–1065.
14. Buchman T, Stein P, Goldstein B (2002) Heart rate variability in critical illness and critical care.
Curr Opin Crit Care 8: 311–315.
10
15. Kleiger RE, Stein PK, Bigger JT (2005) Heart rate variability: Measurement and clinical utility.
Annals of Noninvasive Electrocardiology 10: 88–101.
16. Tsuji H, Larson MG, Venditti FJ, Manders ES, Evans JC, et al. (1996) Impact of reduced heart
rate variability on risk for cardiac events: The framingham heart study. Circulation 94: 2850–2855.
17. La Rovere MT, Pinna GD, Maestri R, Mortara A, Capomolla S, et al. (2003) Short-term heart
rate variability strongly predicts sudden cardiac death in chronic heart failure patients. Circulation
107: 565–570.
18. MacArthur Research Network on SES & Health (2008) Heart Rate Variability. University of
California San Francisco, http://www.macses.ucsf.edu/research/allostatic/.
19. Umetani K, Singer DH, McCraty R, Atkinson M (1998) Twenty-four hour time domain heart rate
variability and heart rate: Relations to age and gender over nine decades. Journal of the American
College of Cardiology 31: 593–601.
20. Bansal D, Khan M, Salhan AK (2009) A review of measurement and analysis of heart rate variabil-
ity. In: Proc. International Conference on Computer and Automation Engineering, March 8–10,
2009, Bangkok, Thailand, pp. 243–246.
21. Laguna P, Caminal P, Jane R, Rix H (1990) Evaluation of HRV by PP and RR interval analysis
using a new time delay estimate. In: Proc. IEEE Computers in Cardiology, Sep 23-26, 1990,
Chicago, USA, pp. 63–66.
22. Elgendi M (2012) On the analysis of fingertip photoplethysmogram signals. Current Cardiology
Reviews 8: 14–25.
11
Tables
Table 1. HRV Statistical Variables.
Variable Statistical measurementMAX-MIN Difference between shortest and longest aa intervalSDNN Standard deviation of all aa intervalsRMSSD Root mean square of the difference of successive aa intervalsSDSD Standard deviation of differences between adjacent aa intervals
12
Figures
-6
-4
-2
0
2
4
6
8x 10
-5
mV
15.5 16 16.5 17 17.5-6
-4
-2
0
2
4
6
8
10x 10
-3
time
mV
Figure 1. Two successive beats in (a) fingertip photoplethysmogram (PPG) signal (b)second derivative wave of photoplethysmogram (APG) signal.
13
15.9 16 16.1 16.2 16.3 16.4 16.5 16.6 16.7 16.8-6
-4
-2
0
2
4
6
8
10x 10
-3
time
mV
-6
-4
-2
0
2
4
6
8
mV
DicroticNotch
DiastolicPeak
(a)
(b)
SystolicPeak
a
b
c
d
e
Figure 2. Fingertip photoplethysmogram signal measurement [22]. (a) Fingertipphotoplethysmogram. (b) Second derivative wave of photoplethysmogram. The photoplethysmogramwaveform consists of one systolic wave and one diastolic wave, while the second derivativephotoplethysmogram waveform consists of four systolic waves (a, b, c, and d waves) and one diastolicwave (e wave).
14
2 4 6 8 10 12 14 16 18-6
-4
-2
0
2
4
6
8
10x 10
-3
Time(s)
V
2 4 6 8 10 12 14 16 18-6
-4
-2
0
2
4
6
8
10x 10
-3
Time(s)
V
(a)
(b)
Figure 3. An example of PPG recordings for the same volunteer measured (a) during restand (b) after exercise. It is clear that the heart rate after exercise was higher than during rest.
15
16 18 20 22 24 26 28 300.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
y(=(-0.011(x(+(0.434
r(=(-0.39
No(of(Beats(in(20(seconds
SD
NN
((se
c)(-
at(r
est
16 18 20 22 24 26 28 300
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5y(=(-0.022(x(+(0.669r(=(-0.565
No(of(Beats(in(20(seconds
rMS
SD
((se
c)(-
at(r
est
(a)
(b)
Figure 4. Correlation between heart rate and HRV indices. (a) HR and SDNN, (b) HR andrMSSD. It is clear that the rMSSD index is more negatively correlated with HR for APG signalsmeasured at rest than the SDNN index
16
0 0.02 0.04 0.06 0.08 0.1 0.12 0.140
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5y = 3.557 x + -0.027r = 0.894
SDNN (sec) -at rest
rMS
SD
(se
c) -
at r
est
Figure 5. Correlation between SDNN and rMSSD calculated from APG signals for allsubjects measured at rest. It is clear that the SDNN index is highly correlated with rMSSD index.
17
15 20 25 30 35 40 450.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
yi=i-0.004ixi+i0.296ri=i-0.271 i
Agei(years)
SD
NN
i(se
c)i-
Ati
rest
15 20 25 30 35 40 450
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2 yi=i-0.001ixi+i0.091ri=i-0.12 i
Agei(years)
SD
NN
i(se
c)i-
Aft
erie
xerc
isei
(a)
(b)
Figure 6. Correlation between age and SDNN index. (a) Age and SDNN calculated from APGsignals for all subjects measured at rest, (b) age and SDNN calculated from APG signals for all subjectsmeasured after exercise. It is clear that the SDNN index is more negatively correlated with age forAPG signals measured at rest compared to after-exercise measurements.
18
15 20 25 30 35 40 450.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5yi=i-0.004ixi+i0.306
ri=i-0.217
Agei(years)
rMS
SD
i(se
c)i-
Ati
rest
15 20 25 30 35 40 450
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16yi=i-0.001ixi+i0.056
ri=i-0.091
Agei(years)
rMS
SD
i(se
c)i-
Aft
erie
xerc
isei
(a)
(b)
Figure 7. Correlation between age and rMSSD. (a) Age and rMSSD calculated from APGsignals for all subjects measured at rest, (b) age and rMSSD calculated from APG signals for allsubjects measured after exercise. It is clear that the rMSSD index is more negatively correlated withage for APG signals measured at rest compared to after-exercise measurements.