ORIGINAL ARTICLE

Smartphone Applications (Apps) for Heart Rate Measurementin Children: Comparison with Electrocardiography Monitor

Chi-Lin Ho • Yun-Ching Fu • Ming-Chih Lin •

Sheng-Ching Chan • Betau Hwang •

Sheng-Ling Jan

Received: 6 August 2013 / Accepted: 6 November 2013

� Springer Science+Business Media New York 2013

Abstract Heart rate (HR) measurement is essential for

children with abnormal heart beats. The purpose of this

study was to determine whether HR measurement by

smartphone applications (apps) could be a feasible alter-

native to an electrocardiography (ECG) monitor. A total of

40 children, median age of 4.3 years, were studied. Using

four free smartphone apps, pulse rates were measured at the

finger (or toe) and earlobe, and compared with baseline

HRs measured by ECG monitors. Significant correlations

between measured pulse rates and baseline HRs were

found. Both correlation and accuracy rate were higher in

the earlobe group than the finger/toe group. When HR was

\120 beats per min (bpm), the accuracy rates were not

different between the two different measuring sites for each

app (median of 65 vs 76 %). The accuracy rates in the

finger/toe group were significantly lower than those in the

earlobe group for all apps when HR was C120 bpm (27 vs

65 %). There were differences among apps in their abilities

to measure pulse rates. Taking children’s pulse rate from

the earlobe would be more accurate, especially for tachy-

cardia. However, we do not recommend that smartphone

apps should not be used for routine medical use or used as

the sole form of HR measurement because the results of

their accuracy are not good enough.

Keywords Children � Electrocardiography monitor �Heart rate � Pulse rate � Smartphone apps

Introduction

Due to the difficulty in identifying heartbeat abnormalities in

children, monitoring heart rate (HR) is important, particularly

in children with palpitations or tachycardia, which may occur

at home or outside the hospital [12]. Monitoring HR outside

the clinical setting is also necessary for children at risk of

developing bradycardia in the event of permanent pacemaker

dysfunction [2], or tachycardia in early-stage severe entero-

virus 71 infection [4]. However, palpating the pulses or lis-

tening to heart sounds may be challenging for caregivers who

are not trained in these medical skills. Currently, pediatric

patients may have to travel a considerable distance to visit a

clinic or hospital for uncertainty of arrhythmia attack.

Therefore, there is an urgent need for a convenient modality

which would allow parents or caregivers to accurately mea-

sure HR in children outside the clinical setting.

Nowadays, many free software applications (apps) that

utilize the processing capabilities of smartphones are

available which allow users to obtain their HR. These free

HR apps allow users to determine their heartbeats per min

(bpm) within seconds, and they are designed for self-use.

The user needs to put the tip of the finger on the camera

C.-L. Ho � Y.-C. Fu � M.-C. Lin � S.-L. Jan (&)

Department of Pediatrics, Taichung Veterans General Hospital,

160, Sec. 3, Chung-Kang Road, Taichung 40705, Taiwan

e-mail: sljan@vghtc.gov.tw

C.-L. Ho

Department of Pediatrics, Chang Bing Show Chwan Memorial

Hospital, Changhua County, Taiwan

Y.-C. Fu � M.-C. Lin � S.-L. Jan

Institute of Clinical Medicine, School of Medicine, National

Yang-Ming University, Taipei, Taiwan

S.-C. Chan

Department of Nursing, Ta-Jen University, Pingtung County,

Taiwan

B. Hwang

Department of Pediatrics, Zhongxiao Branch, Taipei City

Hospital, Taipei, Taiwan

123

Pediatr Cardiol

DOI 10.1007/s00246-013-0844-8

lens and the light emitting diode (LED) flashlight simul-

taneously. The apps then record the changes in light and

shadow accompanying the arterial pulsation and process

these to obtain the pulse rates. These smartphone apps are

not intended for medical use and to date no research has

been conducted to determine their accuracy for use in

children [3]. The purpose of this study was to determine the

accuracy of a selection of smartphone apps for HR mea-

surement in children with a view to assessing their feasi-

bility for use as an alternative to an electrocardiography

(ECG) monitor.

Methods

A prospective study of children who underwent ECG

monitoring at Taichung Veterans General Hospital

(TCVGH) between September 2012 and March 2013 was

conducted. The study was approved by the hospital’s

Institutional Review Board [IRB TCVGH No. CF13009].

Patients who were premature babies (whose fingers or toes

were too small to cover the camera and LED of the

smartphone simultaneously), older than 18 years of age, in

a critical condition with unstable vital signs or could not

obtain written informed consent were excluded. The

patients’ demographic data and vital signs were collected.

The device we used in this study was the iPhone 4S (Apple

Inc., CA, USA). Using the keyword ‘heart rate’, five free

apps for measuring HR in the order of search results were

downloaded from the App Store and used to measure the

pulse rates at the patient’s finger (or toe) and earlobe. ECG

monitors were connected to the patient’s chest wall with

leads. If the patient’s index finger was too small to cover

the camera lens and the flashlight simultaneously, the big

toe was used instead. Pulse rate was measured three times

at each site and for no more than 20 s each time. The room

temperature was in the range of 20 �C (68 �F) to 25 �C

(77 �F). Because it was difficult for some children to

cooperate and stay sufficiently still during HR measure-

ments, we decided to take the pulse rate for no more than

20 s in this study.

When the pulse rate showed on the smartphone’s screen

consistently, we stopped the timer and recorded the HR

displayed on the ECG monitor and the pulse rate on the

smartphone simultaneously. The mean values of the pulse

rates measured by the apps, the HRs on the ECG monitors,

and the durations of the measurements were recorded. The

definition of accuracy was (1) when HR was B100 bpm the

difference was within 5 bpm, or (2) when HR was

[100 bpm the difference was within 5 % of heartbeats, by

referring to the specifications of bedside monitor (The

Philips SureSigns VM8). According to the definition of

accuracy, patients were divided into two groups: accurate

and inaccurate (including pulse rates that could not be

measured), and statistical analyses were performed.

Patients were also divided into two groups according to the

HR (\120 and C120 bpm) to determine the feasibility of

using these apps to measure tachycardia outside the

hospital.

Statistical Analysis

Categorical variables were summarized as frequencies with

percentages, and continuous variables as mean ± standard

deviation. PASW statistics 18.0 for Windows (SPSS Inc.,

Chicago, IL, USA) was used for the statistical analyses.

The paired t test was used to compare the pulse rates

measured by the apps and the HRs measured by the ECG

monitor. Pearson’s correlation and linear regression ana-

lysis were performed and the scatter plots were drawn.

Independent samples t test was used to determine signifi-

cant variables. Chi squared test was used to compare the

percentage accuracy of the HR measurement between dif-

ferent groups. A pvalue of 0.05 was used for hypothesis

testing.

Results

A total of 126 patients who were admitted to pediatric and

neonatal intensive care units in our hospital (TCVGH) and

underwent ECG monitoring were enrolled in this study. Of

these, 40 (age range from 3 days to 15.1 years, median

4.3 years) were eligible and consented to participate in the

study. The patients’ data and vital signs are summarized in

Table 1. The diseases of these children included cardiac

Table 1 Demographic data and vital signs of patients

Mean ± SD Range

Gender 26M/14F

Age 5.7 ± 5.4 3 days–

15.1 years

Body weight (kg) 23.1 ± 19.9 3.0–72.0

Body length (cm) 104.7 ± 40.6 48–170

Body temperature (�C) 36.6 ± 0.7 35.2–39.2

Respiratory rate (breaths/min) 34 ± 16 14–95

Systolic blood pressure (mmHg) 106 ± 24 58–162

Diastolic blood pressure (mmHg) 59 ± 15 25–92

Pulse pressure (mmHg) 47 ± 13 20–80

SpO2 (%) 97 ± 6 62–100

Capillary refill time (s) 1.3 ± 0.7 1–4

Baseline heart rate (bpm) 112 ± 23 52–158

bpm beats per min, F females, M males, SpO2 oxygen saturation

measured by pulse oximeter

Pediatr Cardiol

123

diseases (bradycardia in one child, carditis in two, cyanotic

congenital heart diseases in three, non-cyanotic congenital

heart diseases in four, idiopathic pulmonary arterial

hypertension in one), infections (pneumonia in four chil-

dren, severe enterovirus 71 infection in two), neurologic

diseases (brain tumor in two children, cerebrovascular

accident in one, encephalopathy or encephalitis in six,

epilepsy in three, head injury in one), newborns with

abdominal tumor in one child, neonatal jaundice in two,

neonatal infection in one, prematurity in one, respiratory

distress in one, and others (malignancy in one child,

nephritis in two, pancreatitis in one). One of the smart-

phone apps crashed frequently and needed to be removed

and re-downloaded from the App store numerous times. At

some point, the app was removed from the store so research

on this app was discontinued. The other four apps were

named apps A through D. Because of copyright issues, we

do not provide the full name of the apps and the devel-

opers. We substituted some letters by ‘x’, and the apps used

in this study were listed as the follows:

App A—Cardioxxxxx (MaxxxPinxx Lxx)

App B—Hexxx Rxxx Moxxxx (Jxxx Lxx)

App C—Hexxx Rxxx – Frxx (Cxx Resxxxxx)

App D—Inxxxxx Hexxx Rxxx (Azxxxx Ixx).

The paired t-test was used to analyze measurements of

pulse rate at two different measuring sites for each app and

the HRs measured by ECG monitors; the p value was plotted

on scatter plots with a regression line for each app, which is

shown in Fig. 1. The results showed significant correlations

between the pulse rates measured by each app and the

baseline HRs measured by ECG monitors (R2 = 0.215–

0.857, p \ 0.001 to p = 0.003), except for app C measuring

at the finger/toe (R2 = 0.071, p = 0.097). In addition, the

correlation was better in the earlobe group than the finger/

toe group for all four apps. Results of the independent

samples t test between the accurate and inaccurate groups

are shown in Table 2. Generally, patients in the accurate

group had higher body weight, greater body length, and

older age than those in the inaccurate group with signifi-

cance. Accuracy of the pulse-rate measurement was higher

in the earlobe group than the finger/toe group, and the Chi

squared test showed significant differences between the two

different measuring sites for apps A and D (p = 0.039 and

p \ 0.001, respectively). With regard to the feasibility of

taking pulse rates of patients with desaturation, two patients

(10 months and 13.6 years, respectively) had cyanotic

Fig. 1 The scattergrams show the paired data of the pulse rates

measured by smartphone app A–D (y axis) and the HRs measured by

ECG monitors (x axis). The regression lines were drawn. app

application, b1 regression slope, ECG electrocardiography, HR heart

rate, PR pulse rate, R2 the coefficient of determination

b

Pediatr Cardiol

123

Ta

ble

2C

om

par

ing

the

mea

nv

alu

eso

fv

aria

ble

sb

etw

een

accu

rate

and

inac

cura

teg

rou

ps

for

each

app

attw

od

iffe

ren

tm

easu

rin

gsi

tes

Ap

ps

Sit

esn

Ag

e

(yea

rs)

BL

(cm

)

BW

(kg

)

BT

(�C

)

RR

(per

min

)

SB

P

(mm

Hg

)

DB

P

(mm

Hg

)

PP

(mm

Hg

)

Sp

O2

(%)

CR

T

(s)

Bas

elin

e

HR

(bp

m)

Mea

sure

d

tim

e(s

)

Acc

ura

cy

rate

(%)

pv

alu

e

Fin

ger

(to

e)

AA

20

8.8

*1

25

.2*

33

.8*

36

.53

01

13

65

*4

99

71

.0*

10

2*

9.8

50

.00

.03

9

I2

02

.6*

84

.1*

12

.4*

36

.53

99

95

3*

46

98

1.6

*1

22

*1

2.7

Ear

lob

e

A2

96

.31

09

.82

6.3

*3

6.6

34

10

96

24

79

81

.31

07

*9

.77

2.5

I1

14

.19

1.1

14

.5*

36

.33

49

85

14

79

51

.21

25

*8

.9

Fin

ger

(to

e)

BA

26

5.9

10

6.4

25

.23

6.5

35

10

86

04

89

71

.21

09

7.4

65

.00

.63

3

I1

45

.41

01

.41

9.1

36

.53

31

04

57

46

99

1.5

11

88

.5

Ear

lob

e

A2

86

.7*

11

1.1

27

.0*

36

.53

31

07

59

48

97

1.3

11

08

.27

0.0

I1

23

.4*

89

.51

4.0

*3

6.4

36

10

45

94

59

81

.31

16

10

.2

Fin

ger

(to

e)

CA

10

9.1

*1

28

.0*

32

.83

6.7

29

10

95

95

09

81

.69

7*

10

.2*

25

.00

.61

7

I3

04

.6*

97

.0*

19

.93

6.4

36

10

55

94

69

71

.21

17

*1

3.6

*

Ear

lob

e

A1

28

.7*

12

7.1

*3

7.2

*3

6.4

29

11

06

14

99

91

.21

01

9.8

*3

0.0

I2

84

.5*

95

.0*

17

.1*

36

.53

61

04

58

47

97

1.3

11

61

2.9

*

Fin

ger

(to

e)

DA

19

7.5

*1

17

.32

8.1

36

.42

91

12

63

49

96

1.2

11

07

.4*

47

.5\

0.0

01

I2

14

.1*

93

.21

8.5

36

.53

91

01

56

46

98

1.4

11

31

0.1

*

Ear

lob

e

A3

46

.3*

10

9.6

*2

5.2

*3

6.5

35

10

96

1*

48

97

1.3

*1

10

8.7

85

.0

I6

2.3

*7

6.7

*1

1.0

*3

6.4

32

91

45

*4

69

81

.0*

12

48

.5

Th

eac

cura

cyra

tes

inea

chap

pb

etw

een

two

dif

fere

nt

mea

suri

ng

site

sw

ere

com

par

edb

yC

hi

squ

ared

test

Aac

cura

teg

rou

p,

Ap

pap

pli

cati

on

,B

Lb

od

yle

ng

th,

bp

mb

eats

per

min

,B

Tb

od

yte

mp

erat

ure

,B

Wb

od

yw

eig

ht,

CR

Tca

pil

lary

refi

llti

me,

DB

Pd

iast

oli

cb

loo

dp

ress

ure

,H

Rh

eart

rate

,

Iin

accu

rate

gro

up

,P

Pp

uls

ep

ress

ure

,R

Rre

spir

ato

ryra

te,

SB

Psy

sto

lic

blo

od

pre

ssu

re,

Sp

O2

ox

yg

ensa

tura

tio

nm

easu

red

by

pu

lse

ox

imet

er

*p\

0.0

5b

yin

dep

end

ent

sam

ple

st

test

Pediatr Cardiol

123

congenital heart diseases and low SpO2 (89 and 62 %,

respectively), and the accuracy of the apps in these patients

was moderately high (75 and 62.5 %, respectively).

Accuracy rates of the different groups, including patients

with an HR \ 120 bpm versus C120 bpm measured at dif-

ferent sites for each app, were compared and are shown in

Table 3. In the group with an HR \ 120 bpm, accuracy rates

were not significantly different between the two different

measuring sites for apps A, B, and C, but app D had a higher

accuracy rate measured at the earlobe than that measured at

the finger/toe (p = 0.016). Accuracy rate dropped in most

apps and for both measuring sites when patients had an

HR C 120 bpm. Accuracy rates of HR measured at the fin-

ger/toe were all below 50 % (17.6–47.1 %) and were sig-

nificantly lower for apps A, B, and D (p = 0.004, 0.041, and

0.049, respectively). Although the accuracy rates of HR

measured at the earlobe were also lower when patients’ HR

was C120 bpm (except app B), the change was not signifi-

cant for all apps (p = 0.096, 0.944, 0.143, and 0.195,

respectively). Accuracy rates of HR measured at the earlobe

were also better (17.6–76.5 %) than those measured at the

finger/toe in patients with an HR C 120 bpm, with signifi-

cant differences for apps A and D (p = 0.037 and 0.006,

respectively).

Discussion

HR measurement is an important issue in pediatric diseases

characterized by tachycardia, such as a tachyarrhythmia

attack in which palpitations and chest pain or discomfort

occur. HR measurements are also essential for monitoring

bradycardia in children with dysfunction of the permanent

pacemaker, or for detecting early signs of sepsis or severe

enterovirus 71 infection when the tachycardia is not propor-

tional to fever [8]. Besides auscultation of the heart sounds,

the ECG monitor is the modern gold standard and most

widely used method of checking HRs. The ECG monitor

detects and amplifies the electrical changes on the skin caused

by depolarization of the myocardium during contraction.

Pulse rate is representative of HR in most people. Pulse rates

can be conventionally checked by palpation of peripheral

arterial pulsation, manually or mechanically [11]. Another

method of detecting the pulse rate is the use of photople-

thysmography, such as a pulse oximeter, to analyze the

rhythmic skin color change accompanying the pulse when

measuring arterial hemoglobin oxygen saturation [10]. The

aforementioned methods cannot easily be performed by

caregivers who need to measure the heart/pulse rates of

children outside the hospital setting.

Smartphone apps employ various methods for measur-

ing HRs. One method is the phonocardiogram, which uses

the microphone of the cellphone to ‘listen to’ and measure

heartbeats, but requires online server analysis or it must

learn how to recognize an individual user’s heart sounds

[5]. Another is photoplethysmography, as mentioned

above, and many HR apps have been developed based on

this technique [9]. According to the HR apps’ guidelines,

the apps are to be used for checking the pulses at the finger,

and are for adult self-use [6]. Thus, when the caregivers try

to use the apps for children, they may find children’s fin-

gers are too small to cover the camera and the flashlight

simultaneously, and sometimes even the toes may also be

too small. As a pulse oximeter can be used at the earlobe in

some circumstances [7], we speculated that the earlobe

may be a good alternative site for checking the pulse rate in

children. The smartphone apps showed different abilities to

take pulse rates, and the decreased ability of smartphone

apps to work well in small children and those with high

HRs was noted in this study. The accuracy rates of HR

measurement were better at the earlobe than at the finger/

toe, especially for tachycardia. This may be due to the

relatively thinner layer of skin at the earlobe, which allows

better penetration by the flashlight. In addition, the earlobe

may have a better photoplethysmographic amplitude

compared with those of the finger and toe [1].

Study Limitations

First of all, pathological HRs in children are usually much

higher and may exceed 180 bpm. This is a preliminary

study for the accuracy of HRs measured by smartphone

apps, but not specified for arrhythmia attack. The patient’s

Table 3 The accuracy rates of measured pulse rates in groups divi-

ded according to the baseline heart rate for each app at two different

measuring sites

Apps Sites Accuracy rate p value

HR \ 120 bpm HR C 120 bpm

A Finger (toe) 69.6 % (16/23) 23.5 % (4/17) 0.004

Earlobe 82.6 % (19/23) 58.8 % (10/17) 0.096

p value 0.300 0.037

B Finger (toe) 78.3 % (18/23) 47.1 % (8/17) 0.041

Earlobe 69.6 % (16/23) 70.6 % (12/17) 0.944

p value 0.502 0.163

C Finger (toe) 30.4 % (7/23) 17.6 % (3/17) 0.356

Earlobe 39.1 % (9/23) 17.6 % (3/17) 0.143

p value 0.536 1.000

D Finger (toe) 60.9 % (14/23) 29.4 % (5/17) 0.049

Earlobe 91.3 % (21/23) 76.5 % (13/17) 0.195

p value 0.016 0.006

The accuracy rates between groups were compared by Chi squared

test

app application, bpm beats per min, HR heart rate

Pediatr Cardiol

123

HRs in this study were all below 180 bpm. In addition,

accuracy may have been different for each app and on

different devices. The device we used in this study was the

iPhone 4S, but caregivers may use other smartphones.

Besides, the ranking of the apps is always changed by the

time in the App Store. We searched for the apps in August

2012. Some developers occasionally updated their apps for

better performance, and others did not. The same app used

on different phone models may show a different ability to

accurately measure pulse rate in the same individual.

Another issue is that the apps in this study were tested by

medical professionals, not by ordinary people. In the

beginning, we planned that if the results were reliable,

further studies regarding the apps tested by caregivers in

the outpatient setting would be conducted. Other variables,

such as the patient’s level of cooperation, movement by the

patient during the measurement, and the caregiver’s ability

to accurately measure the patient’s pulse rate were not

quantified in this study.

Conclusions

For caregivers who want to measure the HRs of their

children outside the hospital setting, free smartphone HR

apps appear to be attractive options, and the earlobe would

be the preferable site for detecting pulse rates. Although

these apps provide children with a comfortable and con-

venient method of measuring HR, and obviate the need to

carry around additional medical equipment, we do not

recommend their use in routine medical practice because

the results of their accuracy are not good enough. They

should not be solely depended on for measurement of HR

in a clinical setting. In this study, we tested smartphone HR

apps that use photoplethysmography to detect the pulse.

Future studies should include other types of apps that use

another method, such as phonocardiography.

Acknowledgments The authors would like to thank Mr. Peter

Wilds, who obtained the degree of Combined Studies (English and

Psychology) from the University of Newcastle-upon-Tyne (UK), for

his assistance in the English-language editing. We are also grateful to

all our colleagues in the Department of Pediatrics, Taichung Veterans

General Hospital, for their help.

Funding source This project was conducted with no specific

financial support.

Conflict of interest The authors and the English-language editor

declare no conflicts of interest.

References

1. Awad AA, Ghobashy MA, Ouda W, Stout RG, Silverman DG,

Shelley KH (2001) Different responses of ear and finger pulse

oximeter wave form to cold pressor test. Anesth Analg

92:1483–1486

2. Bakhtiary F, Dzemali O, Bastanier CK, Moritz A, Kleine P

(2007) Medium-term follow-up and modes of failure following

epicardial pacemaker implantation in young children. Europace

9:94–97

3. Buijink AW, Visser BJ, Marshall L (2013) Medical apps for

smartphones: lack of evidence undermines quality and safety.

Evid Based Med 18:90–92

4. Chang LY (2008) Enterovirus 71 in Taiwan. Pediatr Neonatol

49:103–112

5. Chen YH, Chen HH, Chen TC, Chen LG (2011) Robust heart rate

measurement with phonocardiogram by on-line template extrac-

tion and matching. Conf Proc IEEE Eng Med Biol Soc

2011:1957–1960

6. Gregoski MJ, Mueller M, Vertegel A, Shaporev A, Jackson BB,

Frenzel RM, Sprehn SM, Treiber FA (2012) Development and

validation of a smartphone heart rate acquisition application for

health promotion and wellness telehealth applications. Int J

Telemed Appl 2012:696324

7. Haynes JM (2007) The ear as an alternative site for a pulse

oximeter finger clip sensor. Respir Care 52:727–729

8. Jan SL, Lin SJ, Fu YC, Chi CS, Wang CC, Wei HJ, Chang Y,

Hwang B, Chen PY, Huang FL, Lin MC (2010) Extracorporeal life

support for treatment of children with enterovirus 71 infection-

related cardiopulmonary failure. Intensive Care Med 36:520–527

9. Jonathan E, Leahy M (2010) Investigating a smartphone imaging

unit for photoplethysmography. Physiol Meas 31:N79–N83

10. Kim JM, Arakawa K, Benson KT, Fox DK (1986) Pulse oximetry

and circulatory kinetics associated with pulse volume amplitude

measured by photoelectric plethysmography. Anesth Analg

65:1333–1339

11. Meyer-Sabellek W, Schulte KL, Gotzen R (1990) Non-invasive

ambulatory blood pressure monitoring: technical possibilities and

problems. J Hypertens Suppl 8:S3–S10

12. Sacchetti A, Moyer V, Baricella R, Cameron J, Moakes ME

(1999) Primary cardiac arrhythmias in children. Pediatr Emerg

Care 15:95–98

Pediatr Cardiol

123