Can Wearable Devices Accurately Measure Heart Rate Variability? A Systematic Review

Abstract

Background: A growing number of wearable devices claim to provide accurate,
cheap and easily applicable heart rate variability (HRV) indices. This is mainly accomplished
by using wearable photoplethysmography (PPG) and/or electrocardiography
(ECG), through simple and non-invasive techniques, as a substitute of
the gold standard RR interval estimation through electrocardiogram. Although
the agreement between pulse rate variability (PRV) and HRV has been evaluated
in the literature, the reported results are still inconclusive especially when using
wearable devices.
Aim: The purpose of this systematic review is to investigate if wearable devices
provide a reliable and precise measurement of classic HRV parameters in rest as
well as during exercise.
Materials and methods: A search strategy was implemented to retrieve relevant
articles from MEDLINE and SCOPUS databases, as well as, through internet search.
The 308 articles retrieved were reviewed for further evaluation according to the
predetermined inclusion/exclusion criteria.
Results: Eighteen studies were included. Sixteen of them integrated ECG - HRV
technology and two of them PPG - PRV technology. All of them examined wearable
devices accuracy in RV detection during rest, while only eight of them during
exercise. The correlation between classic ECG derived HRV and the wearable RV
ranged from very good to excellent during rest, yet it declined progressively as
exercise level increased.
Conclusions: Wearable devices may provide a promising alternative solution for
measuring RV. However, more robust studies in non-stationary conditions are
needed using appropriate methodology in terms of number of subjects involved,
acquisition and analysis techniques implied.

Full text

7
Folia Medica I 2018 I Vol. 60 I No. 1
DOI: 10.2478/folmed-2018-0012
REVIEW
Can Wearable Devices Accurately Measure Heart Rate Variability?
A Systematic Review
Konstantinos Georgiou1, Andreas V. Larentzakis2, Nehal N. Khamis3, Ghadah I. Alsuhaibani3,
Yasser A. Alaska3, Elias J. Giallafos4
1 Department of Biological Chemistry, Faculty of Medicine, National and Kapodistrian University of Athens, Athens, Greece
2 1st Propaedeutic Surgical Clinic, Medical School, National and Kapodistrian University of Athens, Athens, Greece
3 Clinical Skills & Simulation Center, Department of Medical Education, College of Medicine, King Saud University, Riyadh, KSA
4 Medical Physics Simulation Center, Medical School, National and Kapodistrian University of Athens, Athens, Greece
Correspondence:
Konstantinos Georgiou, Depart-
ment of Biological Chemistry,
Faculty of Medicine, National and
Kapodistrian University of Athens,
75 Mikras Asias Str., Goudi, 11527
Athens, Greece
E-mail: kongeorgiou@med.uoa.gr
Tel: +306942066216
Received: 27 Dec 2017
Accepted: 19 Jan 2018
Published Online: 31 Jan 2018
Published: 30 March 2018
Key words: heart rate, heart rate
variability, wearable, photople-
thysmography
Citation: Georgiou K, Larentzakis
AV, Khamis NN, Alsuhaibani GI,
Alaska YA, Giallafos EJ. Can wear-
able devices accurately measure
heart rate variability? A system-
atic review. Folia Med (Plovdiv)
2018;60(1):7-20.
doi: 10.2478/folmed-2018-0012
Background: A growing number of wearable devices claim to provide accurate,
cheap and easily applicable heart rate variability (HRV) indices. This is mainly ac-
complished by using wearable photoplethysmography (PPG) and/or electrocar-
diography (ECG), through simple and non-invasive techniques, as a substitute of
the gold standard RR interval estimation through electrocardiogram. Although
the agreement between pulse rate variability (PRV) and HRV has been evaluated
in the literature, the reported results are still inconclusive especially when using
wearable devices.
Aim: The purpose of this systematic review is to investigate if wearable devices
provide a reliable and precise measurement of classic HRV parameters in rest as
well as during exercise.
Materials and methods: A search strategy was implemented to retrieve relevant
articles from MEDLINE and SCOPUS databases, as well as, through internet search.
The 308 articles retrieved were reviewed for further evaluation according to the
predetermined inclusion/exclusion criteria.
Results: Eighteen studies were included. Sixteen of them integrated ECG - HRV
technology and two of them PPG - PRV technology. All of them examined wear-
able devices accuracy in RV detection during rest, while only eight of them during
exercise. The correlation between classic ECG derived HRV and the wearable RV
ranged from very good to excellent during rest, yet it declined progressively as
exercise level increased.
Conclusions: Wearable devices may provide a promising alternative solution for
measuring RV. However, more robust studies in non-stationary conditions are
needed using appropriate methodology in terms of number of subjects involved,
acquisition and analysis techniques implied.
BACKGROUND
Monitoring and analysis of heart rate (HR) pro-
vide valuable information regarding health status
and have been extensively investigated in various
activities of healthy subjects as well as in patients
suffering from various diseases.1,2 Heart rate vari-
ability (HRV) has emerged as a non-invasive tool
to estimate the vagal activity in several conditions,
including monitoring of athletic responses to training.
Decreased HRV has been reported as a predictive
factor for adverse outcomes in disease states and
has been found to be associated with fatigue, stress,
and even burnout during athletic performance.3-5
High HRV is an indication of a better general health
status as it allows better adjustment to external and
internal stimuli.6
Due to the fact that traditional HRV recording
methods, such as using electrocardiography (ECG)
and specialized software, often involve expensive
equipment, which is primarily found in research
laboratories, alternative methods have been used,
yet, with variable results. Photoplethysmography
(PPG) is a simple and low-cost method used to
detect volumetric changes in the peripheral blood
circulation at the skin surface.7
In recent years, several wearable pulse rate moni-

8
K. Georgiou et al
Folia Medica I 2018 I Vol. 60 I No. 1
tors using PPG technology have been developed and
have become widely available. The general concept
of operation of these small, robust and user-friendly
devices is that they contain sensors which reliably
monitor minor changes in the intensity of light from
high intensity light emitting diodes (LEDs) that is
transmitted through or refl ected from the human
tissues. Although they have obvious advantages
over the classical ambulatory ECG recording, the
fact that they use PPG, i.e. a different detecting
approach, raises the question of how much accurate
and reliable are their results when compared to the
gold standard ECG method?
The purpose of this systematic review is to
present the available literature comparing the ECG
derived RR and HRV with that of the wearable
commercially available devices in terms of accuracy
and reliability as well as to reveal their strengths
and limitations in the everyday clinical practice.
MATERIALS AND METHODS
SEARCH STRATEGY
This systematic review was conducted by search-
ing medical literature in MEDLINE and SCOPUS.
The search was guided by the Preferred Reporting
Items for Systematic Reviews and Meta-Analyses
(PRISMA) statement in conjunction with the PRIS-
MA explanation and elaboration document.8,9 The
last search was conducted in April 2017. All the
retrieved article titles and abstracts were screened for
relevant manuscripts. Full texts were then retrieved
for the relevant articles and these were thoroughly
read to judge if they meet the inclusion criteria of
the current systematic review.
Medical Subject Heading (MeSH) terms and text
words were used based on the following search
strategy:
Group A terms: (HRV) OR heart rate variability
Group B terms: ((((smart) OR Smartphone$) OR
wearable$) OR phone$ OR plethysmography OR
photoplethysmography OR impedance)
Group C terms: (((((holter) OR continuous
ECG) OR continuous electrocardiogram) OR con-
tinuous electrocardiography) OR ambulatory ECG)
OR ambulatory electrocardiogram OR ambulatory
electrocardiography.
Group A, group B, and group C were combined
and humans’ studies and English language limits
were applied.
We also searched the reference lists of retrieved
full text review manuscripts for the relevant articles
(regardless of their publishing date) and included
these in our systematic review.
Well defi ned inclusion/exclusion criteria were
applied to fi lter the retrieved literature. Of the ar-
ticles retrieved through the above described search
strategy only those that met the following criteria
were considered for this systematic review:
Inclusion criteria:
Studies related to human subjects.
Studies reported in full text English language.
Studies related to HRV. Studies on HRV detec-
tion and/or analysis and/or interpretation and/or
fi ltering were included.
Studies that compare ECG/Holter recordings of
HRV with any other wearable HRV detection and
capture method(s) were included.
Original papers were only included; however,
reference lists from other kind of manuscripts were
used to retrieve any relevant original studies.
Exclusion criteria:
Studies that did not include HRV were excluded.
St udies that did not use commercially available
wearable hardware were excluded.
In addition, we performed Google search for
wearable devices which claimed that they can
measure HRV via plethysmography HR and are
available on the market. Our search spanned the
last 5 years. The devices list (Table 1) was used as
an additional retrograde search tool for any relevant
studies through manufacturers’ commercial websites.
Two of the authors (KEG and AL) independently
applied the above described search strategy to re-
trieve and screen the articles. Any disagreements
were resolved by a third author (AJG) and a fi nal
decision was made accordingly.
RESULTS
Applying the search strategy described in the methods
section, we retrieved 57 articles from MEDLINE and
269 from SCOPUS. Exclusion of duplicates yielded
308 articles. Of these articles, 272 were excluded
after title and abstract screening according to the
predefi ned criteria. Thirty-six articles were selected
for full text review. Full text reading resulted in
exclusion of further 30 articles, which were found to
be either irrelevant (n=28) or non-comparison papers
(n=2). The remaining six articles were included in
the study. Additionally, 12 relevant peer-reviewed
articles were identifi ed from the reference list of
the reviews and from manufacturers’ commercial
websites and added to the study. So, 18 articles were
fi nally included in this systematic review. The fl ow
diagram of the selection process is shown in Fig. 1.

Wearable Devices Accuracy of HRV Estimation
9Folia Medica I 2018 I Vol. 60 I No. 1
Table 1. Capable to measure HRV wearable models: sensor location, release year and connectivity type
Model Sensor site Company Release Year Connectivity
1 4IIII VIIIIVA chest 4IIII 2013 B & A
2 60beat HR Monitor chest 60beat 2016 B & A
3 HRM Blue chest BlueLeza 2016 B & A
4 Dash Earphones ear Bragi 2014 B
5 TP3 chest Cardiosport 2015 B & A
6 Hexoskin torso Carre Technologies 2014 B & USB
7 Empatica E4 Wristband wrist Empatica 2016 B
8 R2 Smart Fitness HRM Wristband chest Cositea 2016 B & A
9 EQ02 LifeMonitor chest Equivital 2012 B, P & USB
10 Forerunner 935 wrist Garmin 2016 B & A
11 HRM Tri wrist Garmin 2016 B & A
12 Premium Heart Rate Monitor chest Garmin 2016 B & A
13 Vívoactive HR+ wrist Garmin 2015 B & A
14 910XT wrist Garmin 2016 B, P & A
15 920XT wrist Garmin 2016 B, P & A
16 HeartMath emWave Pro ear HeartMath 2015 B & A
17 Athos torso, legs & thighs Mad Apparel 2015 B
18 HxM Smart HR chest Medronic 2017 B & Gateway
19 Alpha 2 wrist Mio 2015 B & A
20 Oxstren hand Oxstren 2015 B
21 H7 Heart Rate Monitor chest Polar 2012 B & WiFi
22 H10 chest Polar 2016 B & WiFi
23 QardioCore chest Qardio 2017 B
24 SmartBand 2 wrist Sony 2016 B
25 Smart Sensor chest Suunto 2017 B & A
26 Spartan Sport wrist Suunto 2017 B & A
27 DGYAO® Mobile HR Monitor ear Top Yao 2016 B
28 Tickr Heart Rate Monitor chest Wahoo Fitness 2015 B & A
29 Tickr X Workout Tracker chest Wahoo Fitness 2017 B & A
30 WHOOP Strap 2.0 wrist Whoop 2016 B
31 Tinké Fitness & Wellness Tracker fi ngers Zensorium 2016 B & A
B: Bluetooth, A: ANT+, P: proprietary

10
K. Georgiou et al
Folia Medica I 2018 I Vol. 60 I No. 1
Figure 1. Flow diagram.
Table 2 summarizes the quantitative and quali-
tative methodology used and the categorization
(measurement, position, location, sample number,
technology used) as well as the main fi ndings of
the included studies.
The total number of subjects involved in the
included studies was 686, with one study hav-
ing 339 subjects.10 All 18 articles have examined
healthy subjects except one: Vasconcellos et al.
(2015) who studied obese adolescents.11 Also, one
study examined children.12 In most of the articles
(n=16), a chest device was used,10-25 while a fi n-
ger device was used in two studies26,27. Sixteen
of the studies utilized a similar to ECG - HRV
technology10-25 and the remaining two integrated
PPG - PRV technologies26,27.
REST
All studies examined the HRV/PRV at rest as
baseline. Overall, agreement between the several
indices of HRV and PPG as measured by Holter
and wearable devices, respectively, was very good
to excellent (ranging from 0.85 to 0.99). The RR
interval correlation ranged from 0.91 to 0.999.
Moreover, in two studies,12,15 the error rate in
detection of R waves was evaluated and found to
range between 0.28 and 0.4%. This was estimated
as an accepted ratio. Regarding time domain indi-
ces of HRV, correlations ranged from 0.98 to 0.99,
while in the frequency domain the correlation was

Wearable Devices Accuracy of HRV Estimation
11Folia Medica I 2018 I Vol. 60 I No. 1
Table 2. Results
# Study Rest Supine Upright
/ Sitting
Walking /
Running /
Exercise
Wearable
device
Location
n
Wearable
device
Technology
Comparison Details
1 Akintola et
al., 2016
Y Y Y Y Chest
(Belt)
18 ECG Correlation
depends on
artifacts
Average artifact % → 19%. AMV
(r: 0.967), SDNN (0.393), RMSSD
(0.285), SDANN (0.680), pnn50
(0.982).
2 Esco & Flatt
2017
Y Y Y N Finger
(silicone
sheath)
30 PPG Good agreement LOA Ln RMSSD constant error ± SD:
-0.13 ± 2.83 for the supine values,
-0.94 ± 3.47 for the seated values,
-1.37 ± 3.56 for the standing values. (r
values from 0.98 to 0.99).
3 Flatt &
Esco, 2013
Y Y N N Chest
(Belt)
25 ECG Total agreement No signifi cant difference, correlation
nearly perfect for RMSSD (r: 0.99).
4 Gamelin et
al, 2006
Y Y Y N Chest
(Belt)
20 ECG Good agreement Supine vs. standing: differences for
uncorrected & corrected RR coeffi cient
correlation: 0.88 & 0.91 for supine &
standing. No differences except RMS-
SD, SD1 in standing. Detection Error
Rate of R waves: 0.4%.
5 Gamelin et
al, 2008
Y Y N N Chest
(Belt)
12 ECG No difference Correlation between ECG & Polar RR
intervals (corrected & uncorrected) was
0.80. No signifi cant differences for
Time Domain, FFT & Poincare plot
except for SD2. R waves detection er-
ror rate: 0.28%.

12
K. Georgiou et al
Folia Medica I 2018 I Vol. 60 I No. 1
6 Giles et al,
2016
Y Y Y N Chest
(Belt)
18 ECG Accurate. Good
agreement
No signifi cant differences for SDNN,
RMSSD, pNN50, VLF, LF, HF, nnLF
7 Heathers,
2013
Y N Y Y Finger
(silicone
sheath)
20 PPG Good agreement
at rest.
Mildly reduced
agreement during
exercise.
Experiment 1: Close agreement – small
overall bias & acceptable limits of
agreement. RR & PP correlation coef-
fi cient (0.988 – 0.999). All HRV from
SPRV > than ECG.
Experiment 2: Close agreement be-
tween RR & PP (highest at rest, r:
0.993 – 0.997, slightly attenuated in
exercise (0.965 – 0.998).
8Hernando et
al, 2016
Y N Y Y Chest (Belt) 23 ECG As exercise
increases, correla-
tion decreases
19/23 had high correlation at rest (r>
0.8). Discrepancy increased from 1.67%
(at rest) → 4.8% at the exercise peak. As
exercise increases, reliability & agree-
ment indices drop below 0.5.
9Hong et al,
2009
N N N Y Chest
(Bioshirt)
18 ECG As exercise
increases, cor-
relation decreases
(artefacts)
Coeffi cient correlation of HRV (r2: 0.965
or higher), HF, LF
10 Kingsley et
al, 2005
Y N Y Y Chest (Belt) 8 ECG High agreement
higher at rest.
Reduced agree-
ment at exercise.
Short relationship between RR internals
during exercise. Similar results obtained
for the RR internal but signifi cant differ-
ences occurred for HRV indices.
11 Nunan et al,
2008
Y Y N N Chest (Belt) 33 ECG Accurate / Good
agreement
No signifi cant differences for HRV indi-
ces – LOA for mean RR, LFnn, HFnn,
LF/HF
12 Nunan et al,
2009
Y Y N N Chest (Belt) 33 ECG Good agreement
Not for all indi-
cators
Correlation coeffi cient: 0.99, 0.86, 0.85
for mean RR, LFnn, HFnn respectfully.
Near perfect correlation for SDNN,
RMSSD (0.99, 0.37). Good correlation
for LF (0.92), HF (0.94), LF/HF (0.87).
All measures of HRV ranged from 0.85-
0.99.

Wearable Devices Accuracy of HRV Estimation
13Folia Medica I 2018 I Vol. 60 I No. 1
LOA: level of agreement
13 Plews et al,
2017
Y N Chest (Belt) 29 ECG Acceptable agree-
ment
Almost perfect correlation all HRV indi-
ces (RMSSD)
14 Romagnoli et
al, 2014
Y N Y Y Chest (Vest) 12 ECG Good agreement
for RR / unequal
results for HRV.
Mean RR, SDNN & SD2, excellent LOA.
RMSSD, HF, LF/HF, HF & SD1 worst
LOA.
15 Vanderlei et
al, 2008
Y N Y Y Chest (Belt) 15 ECG As exercise
increases, correla-
tion decreases.
HRV indices assessed: PNN50, RMSSD,
LFnu, HFnu, LF/HF.
16 Vasconcellos
et al, 2015
Y Y N Chest (Belt) 14 ECG Very good cor-
relation
Moderate to strong agreement (0.68-0.98)
for HR, RR, pNN50, rMSSD, LF, HF,
LF/HF. In general moderate agreement
for frequencies domain measures.
17 Wallen et al,
2012
Y Y N N Chest (Belt) 339 ECG Good to excellent
agreement
328 subjects. Gender & age dependent:
females especially > 60 yrs old present
reduced correlation). ICC coeffi cients ↓ in
women than men → more pronounced in
SDNN.HRV from Polar → Gender + Age
dependent. ICC > 0.8 on all HRV param-
eters among men. Agreement moderate
on all HRV among men.
18 Weippert et
al, 2010
Y Y N Y Chest (Belt) 19 ECG Limited agree-
ment
Good correlation (narrow LOA & small
bias). ICC for HRV frequencies were
high, however in most cases LOA
showed unacceptable discrepancies.

14
K. Georgiou et al
Folia Medica I 2018 I Vol. 60 I No. 1
found to range from 0.85 to 0.94. Finally, in the
non-linear multivariate analysis, the correlation was
found to be > 0.9.
EXERCISE
Eight out of the 18 examined studies used an exer-
cise protocol, with a total of 131 participants. All of
them reported that although there was an excellent
correlation between HRV and PPG as measured by
Holter and wearable devices at rest, this seemed to
decrease up to 0.85 as the level of exercise and/or
motion increased. Overall, RR agreement was moder-
ate to excellent ranging from 0.786 to 1. Regarding
time domain, HRV parameters correlation was found
to range from 0.786 (at the exercise peak) to 1 (rest
and exercise 1st phase). Similar pattern occurred
in the frequency domain HRV parameters where
the correlation was found to range from 0.8 to 1.
Also, RR to pulse pressure (PP) wave correlation
was ranging from 0.8 to 0.998.
DISCUSSION
In recent years, technology advances have been used
to capture HRV through wearable devices during
daily activities. The accuracy of these devices versus
classical methods like ECG is still under evaluation.
This systematic review aims to present the avail-
able relevant literature and discuss their fi ndings,
their limitations as well as to provide possible
explanations for these fi ndings.
It is important to realize that the basic difference
between PPG and ECG is the captured signal per
se: the electrical activity of the heart is depicted
by ECG, whereas the PPG is a mechanical signal
measuring the propagation of the peripheral pulse
wave. Therefore, the time of propagation of the
PP wave from the heart to the distal arterioles is
called pulse transit time (PTT). It is a measure
of the time that elapsed between the R-wave of
QRS complex in the ECG and the arrival point
to PPG device.28 Several studies have shown that
PTT seems to be a surrogate marker of ANS in
parallel to HRV29 and that PTT is dependent on the
properties of the pulse wave velocity, the vascular
path from the heart to the location of the detector
and is negatively correlated with blood pressure,
arterial stiffness and age.7
Therefore, we will discuss HRV and PPG fun-
damentals as well as the factors affecting them
in order to elaborate PPG versus ECG for HRV
measurement.
HRV FUNDAMENTALS
HRV analysis is a widely available and accurate
non-invasive technique used as a quantitative assess-
ment tool of the autonomic nervous system (ANS)
function.30 Studies have shown that reduced HRV
indices, as assessed by the RR interval analysis, is
associated with increased cardiovascular morbidity
and mortality in patients with various diseases as
well as in the general population.31 In addition, heart
rate and HRV analysis have been used for estimation
of mental stress and athlete’s fi tness levels, fatigue
and overload.5
The quantifi cation of ANS function is feasible
by calculating several HRV parameters according
to time-domain, frequency-domain and nonlinear
analysis of consecutive RR intervals.1 These indices
represent different components of the sympathetic
and/or parasympathetic system of the ANS. For
instance, the high frequency (HF) component de-
rived by the frequency domain analysis represents
the parasympathetic activity, while the LF/HF ratio
represents the balance of sympathetic to parasym-
pathetic activity.32
However, it is important to realize that these
HRV indices depends on the recording quality, the
subject’s activity during the recording, the exclu-
sion of artifacts, the detection of arrhythmic beats
and the recording duration ranging from seconds
(short term HRV) to even days (long term). Some
of these indices, like the root mean square of
standard deviation (RMSSD) of RR interval can be
calculated from a 10-second session while others
need more than one-hour recording time.33
Despite being accepted as gold standard methods
for RR interval monitoring and analysis, both the
classical ECG and the ambulatory Holter monitoring
have several drawbacks regarding the proper and
accurate detection of RR intervals. For example,
patients with tremor or elderly patients with fragile
skin have bad quality of recordings with a lot of
noise and artifacts.34 Similarly, other factors such
as surface electromyography, increased electrode
impedance, respiration induced baseline drift, and
electrode contact movement can cause noise and
motion artifacts.7 Additionally, morphological varia-
tions in the ECG waveform and heterogeneity in the
QRS complex can often make diffi cult to identify
the RR interval.35 Another limitation can be the
need for the presence of a specialized technician/
doctor, thus increasing the cost and accordingly
decreasing the wide applicability. Finally, a reported
drawback in ECG wearable devices that do not
record standard ECG derivations is their inability

Wearable Devices Accuracy of HRV Estimation
15Folia Medica I 2018 I Vol. 60 I No. 1
to distinguish some arrhythmias and ectopic beats.10
PPG FUNDAMENTALS
Photoplethysmography (PPG), a cheap, simple
and widespread technology has been used as an
alternative approach to obtain HRV indices.7 The
PPG based devices have a sensor that uses infrared
emitter and a detector. This emitter is integrated
to a probe which is comfortable to wear in stable
places of the body that are rich in microcirculation.
Thus, the blood volume changes in the microvascu-
lar bed which are synchronous to the heartbeat can
be traced without the inconveniences of electrode
installation or the need to undress the examinee.1
The simplicity of the technique, cost-effectiveness,
easy signal acquisition and remote monitoring are
the main and obvious advantages of the PPG versus
the gold standard ECG. Therefore, PPG is often
used in conditions of measurements where mobil-
ity, simplicity, time effi ciency, fl exibility and low
cost are of paramount importance, e.g., in exercise
monitoring, every day motion, monitoring of the
elderly, or with disabled patients, etc.7,36
In the relevant literature of the wearable PPG
devices the terms ‘heart rate’ and ‘pulse rate’ are
frequently used interchangeably. Also, the term
‘pulse rate variability’ (PRV), which is derived
from PPG, has been suggested as a potential ana-
log of HRV.37,38
As PRV is further processed identically to HRV,
the derived parameters can be extracted from both
methods such as the standard deviation of normal
to normal (SDNN) R–R intervals (NN), root mean
square of successive differences between adjacent
NN intervals (RMSSD), proportion of NN50 in
total NN intervals (pNN50), low frequency (LF)
power, high frequency (HF) power and LF/ HF
ratio (LF/HF).
Recently, latest technology smart phones ap-
plications with wearable devices26,27 use PPG for
assessing HRV as an alternative to ECG monitoring
in clinical research. More information is available
in Table 2.
WAVELENGTH USED
PPG uses low intensity infrared or green light, which
are more strongly absorbed by the blood than the
surrounding tissues. It has been shown that 530 nm
light (green) PPG showed higher accuracy of pulse
rate detection than the 645 (red) and 470 (blue) nm
light for monitoring HR.39
MEASUREMENT SITE
A fl at skin surface with rich microvasculature is
required to fi rmly attach the PPG sensor to obtain
an accurate measurement. As such, the usual mea-
surement sites for wearable PPGs are the wrist and
the chest. Most of the wearable devices are placed
on the wrist and considerably fewer of them on
the chest.
However, it is worth noting that there is also a
bunch of quasi-wearable devices attached to either
the ear lobe or the fi nger which compared HRV
to PRV parameters.
The ear is chosen as a measuring site because
it is a natural anchoring point, and it is discrete
since the device can be partially hidden by hair.
Weinschenk et al.28 compared PPG to ECG HRV
measurements in deep breathing test in 343 female
subjects in resting conditions and using appropri-
ate parameters they found an excellent correlation.
Finger tips have also been used as a measuring
site but only in stationary conditions: it has been
shown that HRV derived from fi ngertip PPG had
an excellent correlation to ECG in stationary con-
ditions.38 However, a comparative study of fi nger
derived PRV and HRV in healthy subjects40 using
a stationary equipment found a poor correlation
and suggested that fi nger derived PRV might not
be suitable to substitute ECG derived HRV, as has
been mentioned by other authors, too41.
The use of a smart phone camera as PPG sensor
can provide an acceptable agreement for some HRV
indices when compared with ECG,22 but there are
increasing differences in HRV and HR detection
in the setting of movements or exercise42. This is
because the application algorithm is cancelled in
the fi ltering process of inaccurate signals as well
as due to the diffi culty of stabilizing the fi ngertip
on the camera while exercising.
PARAMETERS INVOLVED IN PPG MEASUREMENTS
Several parameters have to be considered when
interpreting PPG measurements. These include:
1. Motion artifact: Special attention must be ex-
ercised during data PPG acquisition to eliminate
motion-induced artifacts.7 The contact force between
the site and the sensor should be considered as PPG
is vulnerable to such type of artifacts. However,
despite the importance of this factor, we found only
one study in 16 male ischemic patients measuring
the accuracy of a smart phone derived pulse rate
versus ECG, where an excellent correlation was
found at rest, which was slightly deteriorated during
exercise.42
2. Respiration: Since respiration alters the intra-
thoracic pressure and causes blood fl ow variations

16
K. Georgiou et al
Folia Medica I 2018 I Vol. 60 I No. 1
in both the veins the DC component of the PPG
waveform shows minor changes with respiration.43
Thus, it has been shown that the short-term variabil-
ity (RMSSD, SD1, and HF) and LF/HF agreement
between PRV and HRV deteriorates as a result of
the vulnerability to respiratory changes.44
3. Age, gender and environmental factors: Normal
HRV values for various age and gender groups are
still not available in the literature. However, it is
well known that the elderly have increased arterial
stiffness which leads to faster pulse transmission in
the periphery and thus pulse transit time (PTT) dif-
ferences observed between HRV and PRV could be
attributed to aging.45 The reviewed articles, except
one,23 involved young population (mean range from
20.9 to 39.2 years).
Regarding gender infl uence, in our review, just
one study showed that measuring HRV at rest
was age and gender dependent, the correlation
with ECG being lower in women than men and
further decreasing in older women.10 As there is
no strong evidence provided so far that age and/or
gender can play a role in the studied correlations,
further studies are needed to investigate these two
variables in different populations while using ap-
propriate sampling and prospective study design
with a longitudinal follow-up.28
Environmental factor effects such as temperature
was investigated in one of the studies. This study
concluded that ambient temperature could induce a
difference in the short-term variables that refl ect the
parasympathetic activity between HRV and PRV.38
4. Software analysis: Some proprietary software
systems for collection and analysis of RV data
exist like the PPT5 or the IthleteTM software ap-
plication14,26,27 or freely available software may
be used e.g. Kubios (http://www.kubios.com)46.
According to guidelines2, manual editing should
be preferred instead of automated data analysis as
automatic fi lters are known to be unreliable and may
potentially introduce errors. In our review only six
studies used automated analysis only whether the
rest used both manual and automated analysis.
5. Statistical analysis: The Bland-Altman plot must
be used to compare the agreement among a new
measurement technique with a gold standard, as
even a gold standard does not imply to be without
error. This plot allows the identifi cation of any
systematic difference between the measurements.47
In our review only 17 studies used this technique
while four studies did not apply the Bland-Altman
analysis and therefore only the correlation, but
not agreement between the two methods, could be
determined from these publications.
6. Sampling rate: The sampling rate is a matter of
difference between the two approaches. Sampling
rate of PPG is usually 20 Hz much less than that
of ECG which is 125 to 250 Hz. This obviously
implies weaker ability of the PPG devices for events
detection.36
COMPARISON OF PPG VS. ECG FOR HRV MEASUREMENT
There are several studies examining the correlation
between HRV and PRV with inconclusive results.40
This may be due to different experimental settings
or to the absence of standardization of the meth-
ods of analysis used.44 It is worth noting that the
disagreement between the two methods does not
apply to the same extent to all HRV parameters.
Additionally, PPG is susceptible to motion artifacts.
As such, the accuracy of PRV as obtained from PPG
should be interpreted with caution.36
REST
Our search revealed that the comparison studies
performed in stationary conditions have generally
revealed that PRV is a good surrogate of HRV.
This is in line with other studies not involving
wearable devices, which also found an acceptable
agreement between HRV and PRV at sitting and
resting positions.48
PRV becomes stronger at a standing than at a
supine position, as it refl ects the mechanical cou-
pling between respiration and thoracic vasculature
tone. Therefore, when a subject changes his/her
position from supine to upright, even in resting
conditions, a PRV divergence from HRV becomes
apparent. Additionally, HRV indices derived from
PPG data are very sensitive to different factors
including noise, artifacts, stature, atherosclerosis,
location of sensor and sampling rate. It is probably
due to these reasons that some studies comparing
the two methods found differences among normal
healthy subjects7,36,45 as well as in patients40.
EXERCISE
There are many non-stationary situations where
autonomic balance signifi cantly changes like in
stress, during motion or exercise. Unfortunately, in
such situations where PRV would be more useful
as a surrogate measurement of HRV, its clinical
value is questionable: a moderate agreement was
observed in some studies about the factors affecting
the measurements when the subject is exercising or
having mental stress i.e. increased noise produc-

Wearable Devices Accuracy of HRV Estimation
17Folia Medica I 2018 I Vol. 60 I No. 1
tion, contraction of muscles which are in contact
with the sensors, sweating, increased intrathoracic
pressure altering the venous return, increased pe-
ripheral vasoconstriction and the respiratory effort
during exercise.
RR interval variables of Bioshirt ECG were
compared to those from conventional ECG and
found that R-peak detecting capabilities of these
two devices were largely similar. However, as
the level of exercise was increasing, the correla-
tion was decreasing due to artifact production. It
must also be noted that a disadvantage of chest
band wearable devices during intense exercise is
the discomfort that a subject senses as the chest
expands with deep breathing.18
Hernando et al. (2016)17 observed that although
an agreement between the detected R-peaks and
the RR intervals from the Polar wearable and ECG
existed, as the exercise intensity level was increased,
the discrepancy of the RR pairs Bland-Altman plot
also increased. They noticed a good correlation in
some but not in all of the HRV indices, due to
the disagreement of the relative error of the Polar
derived high frequency with that of the ECG, as
the level of the exercise increases.17
Akintola et al. (2016) used a chest wearable
device detecting ECG and HRV and reported enor-
mous amount of artifacts during daily activities in
18 healthy subjects and they concluded that this is
a limitation of the wearable device used.13
In agreement with all of the above, another
study showed that the limits of agreement were
deteriorating as the exercise was intensifi ed, implying
an infl uence of adrenergic input, respiratory effort
and unreliable algorithm detection and recording
RR ability.19
In contrast to the many negative results reported
during intense exercise, there are other research-
ers who reported an overall stronger agreement.49
Also, some other research groups reported that PPG
yielded higher HRV values.36 However, all these
studies involved only a sample size of few subjects.
Unfortunately, most of the fi ndings from our
review showed that the correlation was fading out
as the level of exercise and/ or motion increases.
Furthermore, the data from the reviewed studies
are not able to support an in-depth quantitative
analysis due to the differences in their methodology.
FUTURE DIRECTIONS
As wearable healthcare technology and the research
of light propagation in human tissues are progressing,
it is expected that PPG applications will expand.
For instance, there is a growing interest to remotely
depict PPG through imaging such as contactless
video-photoplethysmography (vPPG)50 or imaging
PPG (IPPG)51.
It is essential to develop advanced wearable
devices with higher accuracy, to minimize motion
artifacts as well as improved algorithms to better
detect and identify errors that may occur during
exercise and higher intensity motion.
Additionally, as the availability of wearable
devices is expanding, more research is obviously
warranted to establish age- and sex-dependent
normal PRV values as well as to standardize
both acquisition protocols and analytical methods
in order to get reliable and accurate results, thus
permitting these methods to become a valid sur-
rogate for HRV parameters.
CONCLUSION
Our systematic review revealed that wearable de-
vices, especially those using PPG, may provide a
promising alternative solution for measuring HRV.
However, it is evident that more robust studies in
non-stationary conditions are needed with appropriate
methodology in terms of number of subjects involved,
acquisition and analysis techniques implied, before
being able to recommend any of the commercially
available devices. Therefore, so far wearable devices
can only be used as a surrogate for HRV at resting
or mild exercise conditions, as their accuracy fades
out with increasing exercise load.
REFERENCES
1. Liu AB, Wu HT, Liu CC, et al. The factors infl uence
compatibility of pulse-pulse intervals with R-R inter-
vals. In: EMBC 2013. Proceedings of 2013 Annual
International Conference of the IEEE: Engineering
in Medicine and Biology Society Piscataway, NJ:
IEEE Service Center. 2013; p. 2068-71.
2. Task force of the European Society of Cardiology
and the North American Society of Pacing and
Electrophysiology. Heart rate variability: Standards
of measurement, physiological interpretation and
clinical use. Circulation 1996;93(5):1043-65.
3. Buccelletti E, Gilardi E, Scaini E, et al. Heart rate
variability and myocardial infarction: systematic
literature review and meta-analysis. Eur Rev Med
Pharmacol Sci 2009;13(4):299-307.
4. La Rovere MT, Pinna GD, Maestri R, et al. Auto-
nomic markers and cardiovascular and arrhythmic
events in heart failure patients: still a place in prog-
nostication? Data from the GISSI-HF Trial. Eur J

18
K. Georgiou et al
Folia Medica I 2018 I Vol. 60 I No. 1
Heart Fail 2012;14(12):1410-9.
5. von Rosenberg W, Chanwimalueang T, Adjei T, et
al. Resolving ambiguities in the LF/HF ratio: LF-
HF scatter plots for the categorization of mental and
physical stress from HRV. Front Physiol 2017;8:360.
6. Lehrer PM, Vaschillo E, Vaschillo B, et al. Heart
rate variability biofeedback increases barorefl ex
gain and peak expiratory fl ow. Psychosom Med
2003;65(5):796-805.
7. Allen J. Photoplethysmography and its application
in clinical physiological measurement. Physiol Meas
2007;28:R1-39.
8. Moher D, Liberati A, Tetzlaff J, et al. Preferred
reporting items for systematic reviews and meta-
analyses: the PRISMA statement. J Clin Epidemiol
2009;62(10):1006-12.
9. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA
statement for reporting systematic reviews and
meta-analyses of studies that evaluate health care
interventions: explanation and elaboration. J Clin
Epidemiol 2009;62:e1-34.
10. Wallén MB, Hasson D, Theorell T, et al. Possibili-
ties and limitations of the Polar RS800 in measur-
ing heart rate variability at rest. Eur J Appl Physiol
2012;112(3):1153-65.
11. Vasconcellos FV, Seabra A, Cunha FA, et al. Heart
rate variability assessment with fi ngertip photople-
thysmography and Polar RS800cx as compared with
electrocardiography in obese adolescents. Blood
Press Monit 2015;20(6):351-60.
12. Gamelin FX, Baquet G, Berthoin S, et al. Validity of
the Polar S810 to measure R-R intervals in children.
Int J Sports Med 2008;29:134-8.
13. Akintola A, van de Pol V, Bimmel D, et al. Com-
parative analysis of the Equivital EQ02 life monitor
with holter ambulatory ECG device for continuous
measurement of ECG, heart rate, and heart rate vari-
ability: A validation study for precision and accuracy.
Front Physiol 2016;7:391.
14. Flatt A, Esco MR. Validity of the Ithlete™ smart
phone application for determining ultra-short-term
heart rate variability. J Hum Kinet 2013;39:85-92.
15. Gamelin FX, Berthoin S, Bosquet L. Validity of the
Polar S810 heart rate monitor to measure R-R inter-
vals at rest. Med Sci Sports Exerc 2006;38:887-93.
16. Giles D, Draper N, Neil W. Validity of the Polar
V800 heart rate monitor to measure RR intervals at
rest. Eur J Appl Physiol 2016;116:563-71.
17. Hernando D, Garatachea N, Almeida R, et al. Valida-
tion of heart rate monitor Polar RS800 for heart rate
variability analysis during exercise. J Strength Cond
Res 2016; doi: 10.1519/JSC.0000000000001662.
18. Hong S, Yang Y, Kim S, et al. Performance study
of the wearable one-lead wireless electrocardio-
graphic monitoring system. Telemed J E Health
2009;15:166-75.
19. Kingsley M, Lewis MJ, Marson RE. Comparison of
Polar 810s and an ambulatory ECG system for RR
interval measurement during progressive exercise.
Int J Sports Med 2005;26:39-44.
20. Nunan D, Jakovljevic DG, Donovan G, et al. Lev-
els of agreement for RR intervals and short-term
heart rate variability obtained from the Polar S810
and an alternative system. Eur J Appl Physiol
2008:103:529-37.
21. Nunan D, Donovan G, Jakovljevic DG, et al. Va-
lidity and reliability of short-term heart-rate vari-
ability from the Polar S810. Med Sci Sports Exerc
2009;41:243-50.
22. Plews DJ, Scott B, Altini M, et al. Comparison of
heart rate variability recording with smart phone
photoplethysmographic, Polar H7 chest strap and
electrocardiogram methods. Int J Sports Physiol
Perform 2017;12(10):1324-8.
23. Romagnoli M, Alis R, Guillen J, et al. A novel de-
vice based on smart textile to control heart’s activ-
ity during exercise. Australas Phys Eng Sci Med
2014;37:377-84.
24. Vanderlei LC, Silva RA, Pastre CM, et al. Com-
parison 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
2008;41:854-9.
25. Weippert M, Kumar M, Kreuzfeld S, et al. Com-
parison 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 2010;109:779-86.
26. Esco MR, Flatt AA, Nakamura FY. Agreement be-
tween a smartphone pulse sensor application and
electrocardiography for determining lnRMSSD. J
Strength Cond Res 2017;31:380-5.
27. Heathers JA. Smartphone-enabled pulse rate vari-
ability: an alternative methodology for the collec-
tion of heart rate variability in psychophysiological
research. Int J Psychophysiol 2013;89:297-304.
28. Weinschenk SW, Beise RD, Lorenz, J. Heart rate
variability (HRV) in deep breathing tests and 5-min
short-term recordings: agreement of ear photople-
thysmography with ECG measurements, in 343
subjects. Eur J Appl Physiol 2016;116:1527-35.
29. Drinnan MJ, Allen J, Murray A. Relation between
heart rate and pulse transit time during paced respi-
ration. Physiol Meas 2001;22(3):425-32.
30. Gisselman AS, D’Amico M, Smoliga JM. Optimiz-
ing inter-session reliability of heart rate variability
- the effects of artifact correction and breathing
type. J Strength Cond Res 2017. doi: 10.1519/

Wearable Devices Accuracy of HRV Estimation
19Folia Medica I 2018 I Vol. 60 I No. 1
JSC.0000000000002258.
31. Maheshwari A, Norby FL, Soliman EZ, et al. Low
heart rate variability in a 2-minute electrocardiogram
recording is associated with an increased risk of
sudden cardiac death in the general population: the
atherosclerosis risk in communities study. PLos One
2016;11(8):e0161648.
32. Georgiou K, Larentzakis A, Papavassiliou A. Sur-
geons and surgical trainees acute stress in real op-
erations or simulation: a systematic review. Surgeon
2017;15(6):355-65.
33. Shaffer F, Ginsberg JP. An overview of heart rate
variability metrics and norms. Front Public Health
2017;5:258.
34. Pérez-Riera AR, Barbosa-Barros R, Daminello-Rai-
mundo R, et al. Main artifacts in electrocardiography.
Ann Noninvasive Electrocardiol 2017; doi:10.1111/
anec.12494.
35. Gambarotta N, Aletti F, Baselli G, et al. A review
of methods for the signal quality assessment to
improve reliability of heart rate and blood pres-
sures derived parameters. Med Biol Eng Comput
2016;54:1025-35.
36. Schäfer A, Vagedes J. How accurate is pulse rate
variability as an estimate of heart rate variability? A
review on studies comparing photoplethysmographic
technology with an electrocardiogram. Int J Cardiol
2013;166:15-29
37. Selvaraj N, Jaryal A, Santhosh J, et al. Assessment
of heart rate variability derived from finger-tip
photoplethysmography as compared to electrocar-
diography. J Med Eng Technol 2008;32(6):479-84.
38. Shin H. Ambient temperature effect on pulse rate
variability as an alternative to heart rate vari-
ability in young adult. J Clin Monit Comput
2016;30(6):939-48.
39. Lee J, Matsumura K, Yamakoshi K, et al. Com-
parison between red, green and blue light refl ection
photoplethysmography for heart rate monitoring
during motion. Conf Proc IEEE Eng Med Biol Soc
2013;2013:1724-7.
40. Wong JS, Lu WA, Wu KT, et al. A comparative
study of pulse rate variability and heart rate vari-
ability in healthy subjects. J Clin Monit Comput
2012;26(2):107-14.
41. Pinheiro N, Couceiro R, Henriques J, et al. Can PPG
be used for HRV analysis? Conf Proc IEEE Eng Med
Biol Soc 2016;2016:2945-9.
42. Lee ES, Lee JS, Joo MC, et al. Accuracy of heart rate
measurement using smart phones during treadmill
exercise in male patients with ischemic heart disease.
Ann Rehabil Med 2017;41(1):129-37.
43. Sun Y, Thakor N. Photoplethysmography revisited:
from contact to noncontact, from point to imaging.
IEEE Trans Biomed Eng 2016;63(3):463-77.
44. Chen X, Huang Y, Yun F, et al. Effect of changes in
sympathovagal balance on the accuracy of heart rate
variability obtained from photoplethysmography.
Exp Ther Med 2015;10(6):2311-8.
45. Gil E, Orini M, Bailón R, et al. Photoplethysmog-
raphy pulse rate variability as a surrogate measure-
ment of heart rate variability during non-stationary
conditions. Physiol Meas 2010;31(9):1271-90.
46. Tarvainen MP, Niskanen JP, Lipponen JA, et
al. Kubios HRV–heart rate variability analysis
software. Comput Methods Programs Biomed
2014;113(1):210-20.
47. Bland JM, Altman DG. Statistical methods for as-
sessing agreement between two methods of clinical
measurement. Lancet 1986;1:307-10.
48. Qiu Y, Cai Y, Zhu Y, et al. Poincaré plot analysis for
pulse interval extracted from non-contact photople-
thysmography. Conf Proc IEEE Eng Med Biol Soc
2005;2:1964-7.
49. Chreiteh S, Belhage B, Hoppe K, et al. Sternal pulse
rate variability compared with heart rate variability
on healthy subjects. Conf Proc IEEE Eng Med Biol
Soc 2014;2014:3394-7.
50. Blackford EB, Estep JR. Effects of frame rate and
image resolution on pulse rate measured using
multiple camera imaging photoplethysmography.
In: B. Gimi, & R. C. Molthen, (Eds.). Proceedings
of SPIE 9417, Medical imaging 2015: Biomedical
applications in molecular structural and functional
imaging. 2015. p. 1-14.
51. Iozzia L, Cerina L, Mainardi L. Relationships be-
tween heart-rate variability and pulse-rate variability
obtained from video-PPG signal using ZCA. Physiol
Meas 2016;37(11):1934-44.

20
K. Georgiou et al
Folia Medica I 2018 I Vol. 60 I No. 1
Возможности переносных приборов для точного измерения вари-
абельности сердечного ритма. Систематическое обозрение
Константинос Георгиу1, Андреас В. Ларенцакис2, Нехал Н. Камис3, Гада И. Алсунхаибани3,
Ясер А. Аласка3, Елиас Дж. Джиалафос4
1 Кафедра биохимии, Медицинский факультет, Национальный Афинский университет им. Каподистрия, Афины, Греция
2 Первая клиника хирургической пропедевтики, Медицинский факультет, Национальный Афинский университет им.
Каподистрия, Афины, Греция
3 Центр клинических умений и симуляции, Медицинский факультет, Медицинский колледж, Университет короля Сауда,
Эр-Рияд, Саудовская Аравия
4 Симуляционный центр медицинской физики, Медицинский факультет, Национальный Афинский университет им. Капо-
дистрия, Афины, Греция
Адрес для корреспонденции:
Константинос Георгиу, Кафедра
биохимии, Медицинский фа-
культет, Национальный Афин-
ский университет им. Капо-
дистрия, ул. „Микрас Асиас” 75,
Гоуди, 11527 Афины, Греция
E-mail: kongeorgiou@med.uoa.gr
Тел: +306942066216
Дата получения: 27 декабря
2017
Дата приемки: 19 января 2018
Дата онлайн публикации: 31
января 2018
Дата публикации: 30 марта
2018
Ключевые слова: сердечный
ритм, вариабельность сер-
дечного ритма, переносные,
фотоплетизмография
Образец цитирования:
Georgiou K, Larentzakis AV,
Khamis NN, Alsuhaibani GI,
Alaska YA, Giallafos EJ. Can
wearable devices accurately
measure heart rate variability?
A systematic review. Folia Med
(Plovdiv) 2018;60(1):7-20
doi: 10.2478/folmed-2018-0012
Введение: Утверждается, что всё большее количество переносных приборов
обеспечивает установление точных, недорогих и легко применимых показате-
лей вариабельности сердечного ритма (ВСР). В основном это обеспечивается
приборами переносной фотоплетизмографии (ФПГ) и/или электрокардиогра-
фии (ЭКГ) с помощью простых и неинвазивных методов в качестве заместите-
лей золотого стандарта оценки интервала RR с помощью электрокардиограм-
мы. Хотя соответствие между вариабельностью частоты пульса (ВЧП) и ВСР
было исследовано в литературе, полученные результаты по-прежнему неубе-
дительны, особенно в отношении использования переносных устройств.
Цель: Целью настоящего систематического обозрения является установление
возможностей переносных приборов обеспечивать надёжное и точное изме-
рение классических параметров ВСР как в состоянии покоя, так и во время
физической нагрузки.
Материалы и методы: Была использована стратегия сбора данных для на-
хождения соответствующих статей в базах данных MEDLINE и SCOPUS, а также
в Интернете. Найденные 308 статей были рассмотрены для дальнейшей оцен-
ки в соответствии с заранее определёнными критериями включения / исклю-
чения.
Результаты: Восемнадцать исследований были включены. В шестнадцати
из них применялась технология ЭКГ – ВСР, а в двух из них - технология ФПГ
– ВЧП. Все они исследовали точность переносных приборов при измерении
СР во время отдыха, а только восемь из них во время физической нагрузки.
Корреляция между классической ВСР, полученной от ЭКГ, и переносной ВСР
варьировалась от очень хорошей до превосходной во время отдыха, но она
постепенно снижалась по мере увеличения нагрузки.
Заключение: Переносные приборы могут обеспечить альтернативное реше-
ние для измерения ВСР. Тем не менее, необходимы более надёжные исследо-
вания в нестационарных условиях с использованием соответствующей ме-
тодологии в отношении количества участвующих в исследовании субъектов,
использованных методов измерения и анализа.