senso s
A icle
Validi y o he Pola H7 Hea Ra e Senso o Hea Ra e Va iabili y
Analysis du ing Exe cise in Di e en Age, Body Composi ion
and Fi ness Le el G oups
Ad ián He nández-Vicen e 1,2,3,* , Da id He nando 4,5 , Jo ge Ma ín-Puyal o 1,2,3, Ge mán Vicen e-Rod íguez 1,2,3,6,7 ,
Nu ia Ga a achea 1,2,3,6,7, Es he Pueyo 4,5,† and Raquel Bailón4,5,†


Ci a ion: He nández-Vicen e, A.;
He nando, D.; Ma ín-Puyal o, J.;
Vicen e-Rod íguez, G.; Ga a achea, N.;
Pueyo, E.; Bailón, R. Validi y o he
Pola H7 Hea Ra e Senso o Hea
Ra e Va iabili y Analysis du ing
Exe cise in Di e en Age, Body
Composi ion and Fi ness Le el
G oups. Senso s 2021,21, 902.
h ps://doi.o g/10.3390/s21030902
Academic Edi o : Ma co Al ini
Recei ed: 22 Decembe 2020
Accep ed: 26 Janua y 2021
Published: 29 Janua y 2021
Publishe ’s No e: MDPI s ays neu al
wi h ega d o ju isdic ional claims in
published maps and ins i u ional a il-
ia ions.
Copy igh : © 2021 by he au ho s.
Licensee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and
condi ions o he C ea i e Commons
A ibu ion (CC BY) license (h ps://
c ea i ecommons.o g/licenses/by/
4.0/).
1G ow h, Exe cise, NU i ion and De elopmen (GENUD) Resea ch G oup, Uni e si y o Za agoza,
50009 Za agoza, Spain; jma inp@uniza .es (J.M.-P.); ge icen@uniza .es (G.V.-R.); nuga a a@uniza .es (N.G.)
2Depa men o Physia y and Nu sing, Facul y o Heal h and Spo Science (FCSD), Uni e si y o Za agoza,
22002 Huesca, Spain
3Red Española de In es igación en Eje cicio Físico y Salud en Poblaciones Especiales (EXERNET), Spain
4BSICOS, A agón Ins i u e o Enginee ing Resea ch (I3A), IIS A agón, Uni e si y o Za agoza,
50015 Za agoza, Spain; dhe nand@uniza .es (D.H.); epueyo@uniza .es (E.P.); bailon@uniza .es (R.B.)
5CIBER de Bioingenie ía, Bioma e iales y Nanomedicina (CIBER-BBN), 50009 Za agoza, Spain
6Cen o de In es igación Biomédica en Red de Fisiopa ología de la Obesidad y Nu ición (CIBER-Obn),
28029 Mad id, Spain
7Ins i u o Ag oalimen a io de A agón-IA2- CITA-Uni e sidad de Za agoza), 50013 Za agoza, Spain
*Co espondence: ahe nandez@uniza .es
† These au ho s con ibu ed equally o he wo k.
Abs ac :
This wo k aims o alida e he Pola H7 hea a e (HR) senso o hea a e a iabili y
(HRV) analysis a es and du ing a ious exe cise in ensi ies in a coho o male olun ee s wi h
di e en age, body composi ion and i ness le el. Clus e analysis was ca ied ou o e alua e how
hese pheno ypic cha ac e is ics in luenced HR and HRV measu emen s. Fo his pu pose, six y-
se en olun ee s pe o med a es consis ing o he ollowing consecu i e segmen s: si ing es , h ee
submaximal exe cise in ensi ies in cycle-e gome e and si ing eco e y. The ag eemen be ween
HRV indices de i ed om Pola H7 and a simul aneous elec oca diog am (ECG) was assessed
using conco dance co ela ion coe icien (CCC). The pe cen age o subjec s no eaching excellen
ag eemen (CCC > 0.90) was highe o high- equency powe (P
HF
) han o low- equency powe
(P
LF
) o HRV and inc eased wi h exe cise in ensi y. A clus e o un i and no young olun ee s
wi h high unk a pe cen age showed he highes e o in HRV indices. This s udy indica es ha
Pola H7 and ECG we e in e changeable a es . Du ing exe cise, HR and P
LF
showed excellen
ag eemen be ween de ices. Howe e , du ing he highes exe cise in ensi y, CCC o P
HF
was lowe
han 0.90 in as many as 60% o he olun ee s. Du ing eco e y, HR bu no HRV measu emen s we e
accu a e. As a conclusion, pheno ypic di e ences be ween subjec s can ep esen one o he causes
o disag eemen be ween HR senso s and ECG de ices, which should be conside ed speci ically
when using Pola H7 and, gene ally, in he alida ion o any HR senso o HRV analysis.
Keywo ds: elec oca diog aphy; wea able de ices; HRV analysis; clus e analysis; exe cise es
1. In oduc ion
Hea a e (HR) a iabili y (HRV) is he oscilla ion in he in e als be ween consecu i e
hea bea s (RR in e als) [
1
]. In he las decades, he use o HRV has been popula ized,
since i allows assessing ca diac au onomic modula ion using simple and non-in asi e
echniques. In gene al, lowe HRV has been associa ed wi h poo e p ognosis in di e en
clinical condi ions, while highe HRV, especially ega ding high- equency oscilla ions, has
been associa ed wi h be e heal h. In pa icula , educed HRV has been epo ed in se e al
ca dio ascula diseases and has been used o isk s a i ica ion, con i ming i s alue as a
Senso s 2021,21, 902. h ps://doi.o g/10.3390/s21030902 h ps://www.mdpi.com/jou nal/senso s
Senso s 2021,21, 902 2 o 14
p edic o o o al and ca diac mo ali y [
2
–
4
]. Also, low HRV has been desc ibed in a wide
ange o non-ca dio ascula diseases, including psychia ic diso de s such as dep ession,
anxie y o schizoph enia [5].
En i onmen al and beha io al ac o s in luence HRV [
6
]. In psychology, HRV is com-
monly used due o i s modula ion by mood s a es, emo ions o cogni i e capaci y [
5
].
Likewise, HRV is a use ul measu emen in he spo s ield since i is sensi i e o changes in
i ness, a igue and pe o mance [
7
]. Heal hy habi s like exe cise, balanced die , mind ul-
ness o psychological in e en ions ha e been shown o inc ease HRV measu es indexing
agal unc ion [
5
,
6
]. Addi ionally, non-modi iable ac o s, such as age o sex, also in luence
he au onomic egula ion o he hea , wi h HRV ha ing been epo ed o p og essi ely
decline wi h age and o p esen enhanced high- equency oscilla ions and a enua ed low-
equency oscilla ions in emales as compa ed o males [
8
]. Because o hese high in e -
and in a-indi idual a ia ions in HRV, i is key o use wea able de ices alida ed o
HRV analysis ha can allow o p ecise in e p e a ion o ca diac esponses o di e en
au onomic s a es.
A ecen sys ema ic e iew and me a-analysis showed ha HRV measu emen s de-
i ed om po able de ices a e gene ally accu a e when compa ed o lab-based elec o-
ca diog am (ECG) [
9
]. Gi en he low cos o HR moni o s i is no su p ising ha hey
a e widely used by p ac i ione s and esea che s. Pa icula ly, Pola Elec o Oy (Kempele,
Finland) is one o he mos well-es ablished b ands in HR moni o ing, wi h Pola H7/H10
HR senso s ha ing been alida ed bo h a es and du ing exe cise [
10
–
12
]. Ne e heless,
p e ious Pola alida ion s udies ha e been ca ied ou in small g oups o young, lean,
heal hy and physically i olun ee s [
10
–
12
]. Howe e , he de ice is commonly used by
indi iduals wi h a ious pheno ypic cha ac e is ics, ega dless o how hese may a ec he
accu acy o he measu emen s [
13
,
14
]. In he me a-analysis desc ibed in [
9
], he absolu e
e o o po able de ices was ound o a y wi h he e alua ed HRV me ic, il / eco e y
posi ion and he pe cen age o women in he s udy sample. The cha ac e is ics o he
subjec s and hei in luence on he measu emen s p o ided by po able de ices ha e no
been analyzed ye .
The ECG eco ds he elec ical ac i i y o he hea using elec odes placed on he
su ace o he body. The e o e, di e ences be ween measu emen s om HR senso s and
ECG could a y depending on he cha ac e is ics o he popula ion unde s udy. To s a
wi h, he age- ela ed myoca dial ib osis p esen in he ca diac issue, he amoun o
subcu aneous a o he elec ode placemen a e expec ed o a ec he ol age acings [
15
].
Addi ionally, ol age will be in luenced by en icula size o mass, as obse ed when
compa ing ained a hle es wi h non-a hle es [
16
]. Acco dingly, some g oups o subjec s
such as men, a hle es o black/A ican ha e been epo ed o ha e highe QRS ol age
and, consequen ly, RR in e als become easie o be de ec ed [
16
]. On he o he hand,
obesi y, olde age and seden a y li es yle may cause lowe ol age, which may esul in
lowe accu acy o po able de ices. In pa icula , unde such ci cums ances, some hea
bea s can be misde ec ed and, while his may no conside ably a ec mean HR, i may
no ably hampe HRV assessmen . Fo hese easons and in ligh o p e ious s udies, we
hypo hesized ha when he quali y o he Pola H7 ECG is comp omised by high noise
du ing in ense exe cise and/o by speci ic pheno ypic cha ac e is ics o he subjec s, HRV
measu es can be dis o ed, pa icula ly hose ela ed o high- equency powe [17].
The pu pose o he p esen s udy was o e alua e he alidi y o HRV analysis de i ed
om RR in e als eco ded by Pola H7 HR senso a es and du ing exe cise and eco e y
in di e en pheno ype g oups based on age, body composi ion and i ness le el.
2. Ma e ials and Me hods
2.1. Subjec s
A o al o six y-se en males ag eed o pa icipa e in he s udy. The sample consis ed
o h ee g oups o olun ee s: 22 young adul s (20–30 yea s old), 22 middle-aged adul s
(40–50 yea s old) and 23 olde adul s (60–70 yea s old). Only subjec s wi hin he p ede-
Senso s 2021,21, 902 3 o 14
ined age anges we e included in he s udy. Subjec s we e excluded om he s udy i
hey we e going h ough an acu e disease, we e su e ing om hea diseases (e.g., hea
ailu e o a ial ib illa ion), we e on ca diac medica ion o p esen ed any clinical condi ion
con aindica ing physical exe cise. Howe e , subjec s who we e o e weigh , seden a y o
su e ing om ch onic diseases such as hype ension, diabe es o hype choles e olemia
we e included in he s udy, because o hei high p e alence in he socie y. Table 1shows
he desc ip i e cha ac e is ics o he h ee age g oups. The s udy was app o ed by he
e hical commi ee o clinical esea ch o A agón (ID o he app o al: PI17/0409), and was
conduc ed by adhe ing o he Decla a ion o Helsinki. A e a clea explana ion o he
po en ial isks o he s udy, all olun ee s p o ided w i en in o med consen .
Table 1. Desc ip i e cha ac e is ics o he h ee age g oups.
Ou come Young Adul s
(n= 22)
Middle-Aged Adul s
(n= 22)
Olde Adul s
(n= 23)
Age (yea s) 25.46 ±2.85 43.17 ±3.32 63.97 ±2.79
Heigh (m) 1.75 ±0.06 1.77 ±0.06 1.71 ±0.05
Weigh (kg) 72.01 ±11.92 78.19 ±10.30 76.31 ±7.76
BMI (m/kg2)23.43 ±2.95 25.02 ±2.83 26.21 ±2.84
Body a (%) 15.25 ±5.59 19.69 ±5.65 23.38 ±5.17
T unk a (%) 16.31 ±6.32 21.29 ±6.38 25.69 ±6.42
PWC80% (W/kg) 2.00 ±0.64 2.01 ±0.58 1.73 ±0.65
Values a e exp essed as mean
±
s anda d de ia ion (SD). BMI = Body mass index; PWC
80%
= Physical Wo k
Capaci y a 80% o maximum HR (208 −0.7 ∗age in yea s) in wa s/kg bodyweigh .
2.2. P ocedu e
All subjec s comple ed one es session. P io o he es , hey we e asked o adhe e o
he ollowing ins uc ions [
18
]: (1) a oid exe cise o s enuous physical ac i i y he day
be o e he es ; (2) d ink plen y o luids o e he 24-h pe iod p eceding he es ; (3) ge
an adequa e amoun o sleep (6–8 h) he nigh be o e he es ; (4) a oid subs ances such
as obacco, alcohol o s imulan s (ca eine, heine, au ine, e c.) in he 8 h be o e he es ;
(5) a oid ood in ake o 3 h p io o pe o ming he es ; and (6) wea com o able, loose-
i ing clo hing. Subjec s’ skin was p epa ed by using a azo o emo e any hai om
he elec ode si es, cleaning he skin wi h alcohol and d ying i wi h a gauze. A 12-lead
high- esolu ion Hol e ECG was acqui ed, wi h he 10 elec odes placed as indica ed by
he manu ac u e (H12+, Mo a a Ins umen , Milwaukee, WI, USA), ensu ing ha hey
did no in e e e wi h he HR senso s ap (Pola H7, Pola Elec o Oy).
The es was conduc ed in an en i onmen ally con olled oom (22–23
◦
C), be ween
16:00–20:00, and was di ided in o 3 consecu i e segmen s: es ing (S
REST
), cycling (S
CY
) and
eco e y (S
REC
). Du ing S
REST
, olun ee s we e moni o ed while sea ed a es o 5 min,
wi hou any mo emen o alking. A pe iod o 2–3 min was es ablished o change om
he chai o he cycle-e gome e , namely om S
REST
o S
CY
, du ing which he subjec ode
he elec ically b aked cycle-e gome e (E goselec 200 K, E goline; Bi z, Ge many) a 50 W
wo kload and chose a cadence which was main ained du ing he en i e es acco ding o
he wo kload and cadence displayed in he cycle-e gome e sc een. S
CY
was a submaximal
cycle-e gome e es di ided in o h ee s ages las ing 5 min each. In o de o a oid a
maximal exe cise es , he maximum hea a e (HRmax) was es ima ed o each subjec by
using he o mula de ined by Tanaka e al. HRmax = 208
−
0.7
∗
age (yea s) [
19
]. Wo kload
was adjus ed du ing each s age o 60, 70 and 80% o HRmax, wi h hese s ages deno ed
as S
CY60
, S
CY70
and S
CY80
, espec i ely. Finally, du ing S
REC
, olun ee s emained sea ed
again o 5 min wi hou any mo emen o alking. Figu e 1shows an example o he
empo al e olu ion o RR in e als om a subjec h oughou he en i e es .
Senso s 2021,21, 902 4 o 14
Senso s 2021, 21, x FOR PEER REVIEW 4 o 14
Wo kload was adjus ed du ing each s age o 60, 70 and 80% o HRmax, wi h hese s ages
deno ed as SCY60, SCY70 and SCY80, espec i ely. Finally, du ing SREC, olun ee s emained
sea ed again o 5 min wi hou any mo emen o alking. Figu e 1 shows an example o
he empo al e olu ion o RR in e als om a subjec h oughou he en i e es .
Figu e 1. Example o he RR in e als o one subjec h oughou he en i e es . Do ed lines sepa-
a e he di e en es segmen s: es ing (SREST), cycling (SCY) and eco e y (SREC). SCY was di ided
in o h ee s ages co esponding o 60, 70 and 80% o HRmax, deno ed as SCY60, SCY70 and SCY80, e-
spec i ely.
2.3. Da a Reco ding
Subjec s sel - epo ed hei bi h da e, cu en diseases and medica ion. The an-
h opome ic cha ac e is ics o he subjec s we e assessed. S a u e was measu ed o he
nea es 0.001 m using a po able s adiome e (SECA 225, Hambu g, Ge many), wi h
subjec s s anding wi h hei scapula, bu ocks and heels es ing agains a wall, he ee
wi h he heels ouching, o ming a 45° angle and he head in he F ank o ’s plane. A
po able body composi ion analyze (TANITA BC-418MA; Tani a Co p., Tokyo, Japan)
was used o measu e he body mass o he nea es 0.1 kg, wi h unde wea and a e u i-
na ion. TANITA BC-418MA was also used o es ima e he pe cen age o body a and
unk a ( = 0.87, p < 0.001 s. dual-ene gy X- ay abso p iome y) [20]. Body mass index
(BMI) was calcula ed di iding weigh in kilog ams by heigh in squa ed me e s.
Bea - o-bea RR in e als wi h 1-ms esolu ion we e ob ained using a Pola V800 HR
moni o simul aneously wi h a Pola H7 ches So S ap (Pola Elec o Oy, hence o h
e e ed o as Pola H7). Concomi an ly, a 12-lead ECG was eco ded a a sampling a e o
1000 Hz using a high- esolu ion Hol e de ice (H12+, Mo a a Ins umen , hence o h
e e ed o as ECG and used he e as a e e ence).
VO2max can be es ima ed om submaximal exe cise es s, a sa e and easible
me hod showing good alidi y agains maximal es s (co ela ion coe icien s: 0.69 o
0.98) [21]. Ra he han commonly used es s wi h s ages o sho o a iable du a ion, an
Figu e 1.
Example o he RR in e als o one subjec h oughou he en i e es . Do ed lines sepa a e
he di e en es segmen s: es ing (S
REST
), cycling (S
CY
) and eco e y (S
REC
). S
CY
was di ided in o h ee
s ages co esponding o 60, 70 and 80% o HRmax, deno ed as SCY60, SCY70 and SCY80, espec i ely.
2.3. Da a Reco ding
Subjec s sel - epo ed hei bi h da e, cu en diseases and medica ion. The an h opo-
me ic cha ac e is ics o he subjec s we e assessed. S a u e was measu ed o he nea es
0.001 m using a po able s adiome e (SECA 225, Hambu g, Ge many), wi h subjec s
s anding wi h hei scapula, bu ocks and heels es ing agains a wall, he ee wi h he
heels ouching, o ming a 45
◦
angle and he head in he F ank o ’s plane. A po able
body composi ion analyze (TANITA BC-418MA; Tani a Co p., Tokyo, Japan) was used
o measu e he body mass o he nea es 0.1 kg, wi h unde wea and a e u ina ion.
TANITA BC-418MA was also used o es ima e he pe cen age o body a and unk a
(
= 0.87
,
p< 0.001
s. dual-ene gy X- ay abso p iome y) [
20
]. Body mass index (BMI) was
calcula ed di iding weigh in kilog ams by heigh in squa ed me e s.
Bea - o-bea RR in e als wi h 1-ms esolu ion we e ob ained using a Pola V800 HR
moni o simul aneously wi h a Pola H7 ches So S ap (Pola Elec o Oy, hence o h
e e ed o as Pola H7). Concomi an ly, a 12-lead ECG was eco ded a a sampling a e
o 1000 Hz using a high- esolu ion Hol e de ice (H12+, Mo a a Ins umen , hence o h
e e ed o as ECG and used he e as a e e ence).
VO
2
max can be es ima ed om submaximal exe cise es s, a sa e and easible me hod
showing good alidi y agains maximal es s (co ela ion coe icien s: 0.69 o 0.98) [
21
].
Ra he han commonly used es s wi h s ages o sho o a iable du a ion, an ad-hoc es
wi h 5-min s ages was de ined o allow eliable es ima ion o he low- equency powe o
HRV. This enabled assessmen o HRV esponse o inc eased sympa he ic ac i i y wi h each
cycling s age [
22
]. Ca dio espi a o y i ness was assessed using he app oach o “Physical
Wo k Capaci y” (PWC). PWC in wa s was measu ed du ing S
CY80
o he submaximal
cycle-e gome e es and was subsequen ly di ided by he subjec ’s body weigh (PWC
80%
in W/kg). Al e na i ely o he use o ixed HR h esholds, his me hod inco po a es he
Senso s 2021,21, 902 5 o 14
age-dependen decline o HRmax [
23
,
24
] and has been p e iously used as an objec i e
assessmen o ca dio espi a o y i ness [25,26].
2.4. Da a Analysis and P ocessing
Raw RR in e al ime se ies, RR
P
(i), eco ded by Pola H7 we e downloaded om he
“Pola Flow” web pla o m. RR in e al ime se ies om he ECG, RRE(i), we e ex ac ed
using a mul i-lead app oach by a wa ele -based de ec o [
27
] wi h op imized pa ame e s
o noisy en i onmen s as desc ibed in [
28
]. Each bea de ec ion was manually e i ied by
an ope a o wi h a dedica ed in e ace.
The delay be ween he RR in e al se ies RR
P
(i) and RR
E
(i) was es ima ed as he ime
lag maximizing hei c oss-co ela ion o e he i s 3 min o he es when he subjec is
elaxed. Then, bo h se ies we e synch onized by compensa ing o his delay. Since he wo
RR in e al se ies can ha e di e en leng hs, due o, e.g., w ong o missed bea de ec ions
in he Pola da a, an algo i hm was de eloped o ma ch he RR in e als om bo h se ies,
hus allowing cha ac e iza ion o he ag eemen be ween he pai ed se ies RR
P
(ip) and
RRE(ip), whe e ip e e s o he indices o bea s ha a e ma ched in he wo se ies.
2.5. Hea Ra e Va iabili y
HRV indices we e ob ained by algo i hms speci ically de eloped and p e iously
published by ou esea ch g oup using MATLAB e sion R2017a (MATLAB, Ma hWo ks
Inc., Na ick, MA, USA) [17,27–30].
2.5.1. Tempo al Domain
The ollowing empo al HRV indices we e s udied [
1
]: mean HR (MHR), s anda d
de ia ion o no mal- o-no mal RR in e als (SDNN) and oo mean squa e o successi e
di e ences o adjacen no mal- o-no mal RR in e als (RMSSD). MHR was ob ained as he
in e se o he mean o he RR in e als. SDNN is conside ed a measu e o he o al powe
o HRV and was calcula ed om he s anda d de ia ion o he NN in e als, i.e., no mal
RR in e als a e co ec ing o ec opic bea s [
29
]. RMSSD is a measu e o sho - e m
a iabili y and was compu ed by he oo mean squa e o successi e di e ences be ween
adjacen NN in e als. These indices we e ob ained om RR
P
(i) and RR
E
(i) in each
segmen o he es .
2.5.2. F equency Domain
The ins an aneous HR signal,
dHR(n)
, was de i ed om bo h RR
P
(i) and RR
E
(i) and
sampled a 4 Hz. The in eg al pulse equency modula ion (IPFM) model was used while
dealing wi h he p esence o ec opic bea s [
29
]. This signal was high-pass- il e ed (0.03 Hz)
o emo e he e y low- equency componen s,
dMHR(n)
, and i was also co ec ed by i :
m(n)= (dHR(n)−dMHR(n))/dMHR(n)[30].
The smoo hed pseudo Wigne –Ville dis ibu ion (SPWVD) was applied o
m(n)
o
es ima e i s ime- a ying spec um. Time and equency smoo hing windows we e chosen
as desc ibed in [
17
]. The ins an aneous powe in he low- equency band, P
LF
(n), was
ex ac ed in eg a ing he SPWVD om 0.04 o 0.15 Hz o each ime ins an . The ins an a-
neous powe in he high- equency band, P
HF
(n), was compu ed in a band cen e ed on
he espi a o y equency wi h a bandwid h o 0.25 Hz. Figu e 2shows an example o
dHR(n)
, P
LF
(n) and P
HF
(n) ob ained om RR
E
. In some analyses, mean P
LF
and P
HF
we e
calcula ed om PLF (n) and PHF (n) o each segmen o he es .

Senso s 2021,21, 902 6 o 14
Senso s 2021, 21, x FOR PEER REVIEW 6 o 14
𝑑(𝑛), PLF (n) and PHF (n) ob ained om RRE. In some analyses, mean PLF and PHF we e
calcula ed om PLF (n) and PHF (n) o each segmen o he es .
Figu e 2. Example o 𝑑(𝑛), PLF (n) and PHF (n) ob ained om RRE o one subjec : Res ing segmen
(le ) and cycling segmen ( igh ). No e ha he axes ha e di e en scales. 𝑑(𝑛)= ins an aneous
HR signal; PLF (n) = Ins an aneous low- equency powe ; PHF (n) = Ins an aneous high- equency
powe ; RRE= RR in e als se ies om he ECG.
2.6. S a is ical Analysis
The no mali y o da a was checked wi h he Kolmogo o -Smi no es . Since he
da a dis ibu ion iola ed he assump ion o no mali y o he pa ame ic es s, and such a
condi ion was no achie ed by commonly employed ans o ma ions, a non-pa ame ic
analysis was pe o med. Desc ip i e alues a e p esen ed as mean ± s anda d de ia ion
(SD) and HRV alues a e epo ed as median and in e qua ile ange. S a is ical analyses
we e pe o med using IBM SPSS ( e sion 25; Chicago, IL, USA). The signi icance le el
was se a p ≤ 0.05.
Wilcoxon es o pai ed samples, he non-pa ame ic equi alen o he pai ed sam-
ples - es , was used o de e mine di e ences be ween he empo al domain HRV da a
ob ained om Pola H7 and om ECG. The magni ude o he di e ences was calcula ed
by de e mining he e ec size (ES): 𝐸𝑆 = 𝑍/√𝑛 whe e Z ep esen s he Z-sco e o he
Wilcoxon s a is ic and n is he o al numbe o obse a ions [31]. Di e ences we e con-
side ed small when ES < 0.2, small o medium when ES = 0.2–0.5, medium o la ge when
ES= 0.5–0.8 and la ge when ES > 0.8 [32].
Lin’s conco dance co ela ion coe icien (CCC) was used o s udy he ag eemen
be ween he ollowing Pola H7-de i ed and ECG-de i ed signals: RR (ip), PHF (n) and PLF
(n). CCC de e mines how much he obse ed da a de ia e om he pe ec conco dance
line a 45° on a squa e axis sca e plo [33]. CCC was e alua ed in each segmen (SREST,
SCY60, SCY70, SCY80 and SREC). A CCC alue g ea e han 0.90 was conside ed “excellen ” [34]
and he pe cen age o subjec s wi h CCC alues below his h eshold was epo ed o
each segmen .
Figu e 2.
Example o
dHR(n)
, P
LF
(n) and P
HF
(n) ob ained om RR
E
o one subjec : Res ing segmen
(
le
) and cycling segmen (
igh
). No e ha he axes ha e di e en scales.
dHR(n)
= ins an aneous
HR signal; P
LF
(n) = Ins an aneous low- equency powe ; P
HF
(n) = Ins an aneous high- equency
powe ; RRE= RR in e als se ies om he ECG.
2.6. S a is ical Analysis
The no mali y o da a was checked wi h he Kolmogo o -Smi no es . Since he
da a dis ibu ion iola ed he assump ion o no mali y o he pa ame ic es s, and such a
condi ion was no achie ed by commonly employed ans o ma ions, a non-pa ame ic
analysis was pe o med. Desc ip i e alues a e p esen ed as mean
±
s anda d de ia ion
(SD) and HRV alues a e epo ed as median and in e qua ile ange. S a is ical analyses
we e pe o med using IBM SPSS ( e sion 25; Chicago, IL, USA). The signi icance le el was
se a p≤0.05.
Wilcoxon es o pai ed samples, he non-pa ame ic equi alen o he pai ed samples
- es , was used o de e mine di e ences be ween he empo al domain HRV da a ob ained
om Pola H7 and om ECG. The magni ude o he di e ences was calcula ed by de e -
mining he e ec size (ES):
ES =Z/√n
whe e Z ep esen s he Z-sco e o he Wilcoxon
s a is ic and nis he o al numbe o obse a ions [
31
]. Di e ences we e conside ed small
when ES < 0.2, small o medium when ES = 0.2–0.5, medium o la ge when ES= 0.5–0.8 and
la ge when ES > 0.8 [32].
Lin’s conco dance co ela ion coe icien (CCC) was used o s udy he ag eemen
be ween he ollowing Pola H7-de i ed and ECG-de i ed signals: RR (ip), P
HF
(n) and
PLF (n)
. CCC de e mines how much he obse ed da a de ia e om he pe ec conco -
dance line a 45
◦
on a squa e axis sca e plo [
33
]. CCC was e alua ed in each segmen
(S
REST
, S
CY60
, S
CY70
, S
CY80
and S
REC
). A CCC alue g ea e han 0.90 was conside ed “ex-
cellen ” [
34
] and he pe cen age o subjec s wi h CCC alues below his h eshold was
epo ed o each segmen .
Clus e analysis was pe o med o iden i y g oups o subjec s wi h simila cha ac e -
is ics in e ms o he ollowing h ee a iables o in e es : age, body composi ion ( unk
a pe cen age) and i ness le el (PWC
80%
). T unk a pe cen age was selec ed among all
body composi ion a iables, since i is he mos speci ic o he elec ode placemen a ea.
Senso s 2021,21, 902 7 o 14
Following he me hodology desc ibed in p e ious s udies [
35
,
36
], wo ypes o clus e anal-
yses we e combined: hie a chical clus e ing (Wa d’s me hod) and k-means clus e ing. Fi s ,
indi idual and mul i a ia e ou lie s (acco ding o Mahalanobis dis ance) we e de ec ed o
educe he sensi i i y o he Wa d’s me hod o ou lie s. Second, hie a chical clus e analysis
was used, as he numbe o clus e s in he da a we e unknown be o ehand. Examina ion o
dend og ams showed ha a ou -clus e solu ion p oduced good di e en ia ion be ween
g oups. Finally, k-means clus e was pe o med wi h ou possible solu ions. Compa ed
o hie a chical me hods, k-means clus e analysis is conside ed less sensi i e o ou lie s
and has been ound o esul in g ea e wi hin-clus e homogenei y and be ween-clus e
he e ogenei y [35].
To assess di e ences in he pe cen age o e o o each HRV index be ween he
ou clus e g oups, a K uskal-Wallis es (non-pa ame ic equi alen o one-way analysis
o a iance, ANOVA) wi h Bon e oni co ec ion was pe o med. The Dunn-Bon e oni
pos hoc me hod was used o pai wise compa isons. The ela i e e o in HRV indices
was calcula ed as he absolu e e o o he Pola H7 wi h espec o he ECG measu emen
di ided by he e e ence ECG measu emen , e.g.,
(SDNNECG −SDNNPola H7)/SDNNECG
,
which was hen mul iplied by 100 o ob ain he pe cen age o e o (%E o ). In he case
o he equency HRV a iables, %E o was calcula ed om he mean alue o each
segmen o he es . To e alua e he magni ude o he di e ences, ES was calcula ed as:
ES =H/(n2−1/(n+1))
, whe e Hs ands o he K uskal-Wallis es s a is ic and nis
he o al numbe o obse a ions [31].
3. Resul s
Table 2shows he desc ip i e cha ac e is ics o he 4 clus e g oups, which we e
desc ibed as CLUSTER A (High PWC
80%
), CLUSTER B (Low PWC
80%
and low age),
CLUSTER C (Low PWC
80%
, high age and medium unk a pe cen age) and CLUSTER D
(Low PWC80%, high age and high unk a pe cen age).
Table 2. Desc ip i e cha ac e is ics o he ou clus e g oups.
Ou come CLUSTER A
(n= 19)
CLUSTER B
(n= 13)
CLUSTER C
(n= 18)
CLUSTER D
(n= 17)
Main E ec
pE ec Size
Age (yea s) 38.99 ±13.31 B,D 24.35 ±2.20 A,C,D 52.61 ±10.20 B57.47 ±12.23 A,B <0.001 * 0.586
Heigh (m) 1.77 ±0.05 D1.72 ±0.06 1.74 ±0.06 1.72 ±0.07 A0.020 0.149
Weigh (kg) 72.65 ±9.13 69.18 ±12.27 D76.32 ±7.13 82.72 ±8.78 B0.007 0.182
BMI (m/kg2) 23.00 ±2.24 D23.23 ±2.77 D25.17 ±1.40 28.04 ±2.79 A,B <0.001 * 0.430
Body a (%) 14.09 ±4.11 C,D 15.45 ±4.54 D
20.40
±
2.36
A,D 27.67 ±2.42 A,B,C <0.001 * 0.743
T unk a (%) 14.76 ±5.11 C,D 16.64 ±5.03 D
22.19
±
2.86
A,D 30.69 ±2.12 A,B,C <0.001 * 0.725
PWC80% (W/kg) 2.73 ±0.39 B,C,D 1.61 ±0.37 A1.79 ±0.28 A,D 1.35 ±0.22 A,C <0.001 * 0.704
Values a e exp essed as mean
±
s anda d de ia ion (SD). BMI = Body mass index; PWC
80%
= Physical Wo k Capaci y a 80% o HRmax
(208
−
0.7
∗
age in yea s) in wa s/kg bodyweigh . Clus e s we e based on: age, body composi ion ( unk a pe cen age) and i ness
le el (PWC
80%
). * = Signi ican di e ences be ween clus e s (p
≤
0.05, K uskal-Wallis es ).
A
= Di e en o CLUSTER A;
B
= Di e en o
CLUSTER B; C= Di e en o CLUSTER C; D= Di e en o CLUSTER D.
Table 3shows he alues o HRV indices ob ained om Pola H7 and ECG. Mean
P
LF
and P
HF
we e calcula ed om P
LF
(n) and P
HF
(n) o each segmen (di e en ly om
Table 4
, whe e he ins an aneous se ies we e used). Wilcoxon es o pai ed samples
e ealed ha P
HF
and empo al domain HRV indices (MHR, SDNN and RMSSD) we e
lowe a all cycling s ages (S
CY60
, S
CY70
and S
CY80
) when measu ed by Pola H7, wi h P
LF
being lowe a he highes in ensi y (S
CY80
) when measu ed by Pola H7. The magni ude o
all hese di e ences was small o medium, i.e., 0.2–0.5 acco ding o he e ec sizes.
Senso s 2021,21, 902 8 o 14
Table 3. HRV indices ob ained om Pola H7 and ECG da a (n= 67).
Pola H7 ECG pES
SREST
PLF (e−4)9.78 (3.88 o 24.74) 9.76 (3.85 o 24.78) 0.074 0.155
PHF (e−4)5.77 (2.30 o 10.91) 5.74 (2.22 o 10.84) 0.067 0.159
MHR (bpm) 62.55 (53.25 o 71.95) 62.84 (53.36 o 72.00) <0.001 * 0.378
SDNN (ms) 60.27 (40.50 o 75.22) 60.28 (40.33 o 75.26) 0.570 0.049
RMSSD (ms) 39.48 (22.48 o 60.04) 39.26 (22.39 o 60.66) 0.336 0.083
SCY60
PLF (e−4)1.17 (0.62 o 2.01) 1.17 (0.65 o 2.23) 0.112 0.137
PHF (e−4)0.51 (0.26 o 1.29) 0.86 (0.44 o 2.07) <0.001 * 0.419
MHR (bpm) 106.00 (100.45 o 112.91) 106.66 (100.75 o 113.17) <0.001 * 0.433
SDNN (ms) 15.94 (12.06 o 20.77) 16.21 (12.85 o 20.32) 0.010 * 0.222
RMSSD (ms) 6.48 (4.34 o 8.51) 8.38 (5.67 o 10.65) <0.001 * 0.482
SCY70
PLF (e−4)0.45 (0.21 o 0.87) 0.46 (0.22 o 0.88) 0.851 0.016
PHF (e−4)0.30 (0.15 o 0.55) 0.41 (0.24 o 0.75) <0.001 * 0.377
MHR (bpm) 124.24 (115.77 o 130.89) 124.32 (115.76 o 130.92) <0.001 * 0.482
SDNN (ms) 10.22 (8.19 o 12.93) 10.48 (8.40 o 13.10) 0.001 * 0.280
RMSSD (ms) 3.74 (2.96 o 4.91) 4.37 (3.64 o 6.68) <0.001 * 0.443
SCY80
PLF (e−4)0.13 (0.09 o 0.23) 0.18 (0.10 o 0.26) <0.001 * 0.364
PHF (e−4)0.23 (0.14 o 0.36) 0.37 (0.25 o 0.66) <0.001 * 0.362
MHR (bpm) 141.01 (130.62 o 148.81) 141.09 (130.69 o 149.05) <0.001 * 0.451
SDNN (ms) 8.11 (6.20 o 9.92) 8.10 (6.34 o 10.65) <0.001 * 0.378
RMSSD (ms) 2.90 (2.32 o 3.90) 3.75 (3.16 o 5.52) <0.001 * 0.405
SREC
PLF (e−4)4.75 (1.83 o 10.25) 4.59 (1.88 o 9.97) 0.881 0.015
PHF (e−4)2.06 (0.65 o 4.71) 2.12 (0.82 o 4.70) 0.308 0.102
MHR (bpm) 99.76 (90.76 o 112.15) 98.39 (90.12 o 111.35) 0.002 * 0.268
SDNN (ms) 33.11 (23.27 o 58.30) 33.67 (23.04 o 57.40) 0.094 0.147
RMSSD (ms) 12.87 (7.65 o 23.78) 12.90 (7.99 o 23.32) 0.603 0.046
Values a e exp essed as median and in e qua ile ange. Segmen s a e based on he es phases: es ing (S
REST
), cycling (S
CY
) and eco e y
(S
REC
). S
CY
was di ided in h ee s ages a 60, 70 and 80% o HRmax, deno ed as S
CY60
, S
CY70
and S
CY80,
espec i ely. P
LF
= low- equency
powe ; P
HF
= high- equency powe ; MHR = mean HR; SDNN = SD o he NN in e als; RMSSD = oo mean squa e o successi e
di e ences be ween NN in e als. ES = E ec size. * = Signi ican di e ences be ween de ices (p
≤
0.05, Wilcoxon es o pai ed samples).
Table 4.
Ag eemen be ween de ices in: RR (ip), PLF (n) and PHF (n). CCC mean and pe cen age o subjec s no eaching
excellen ag eemen o each segmen .
SREST SCY60 SCY70 SCY80 SREC
Whole sample
(n= 67)
RR (ip) 0.9929 (1%) 0.9560 (6%) 0.9467 (13%) 0.9319 (16%) 0.9612 (14%)
PLF (n) 0.9885 (1%) 0.9713 (4%) 0.9677 (9%) 0.9106 (19%) 0.8251 (30%)
PHF (n) 0.9813 (3%) 0.9494 (13%) 0.8858 (27%) 0.6661 (60%) 0.5262 (75%)
CLUSTER A
(n= 19)
RR (ip) 0.9970 (0%) 0.9844 (0%) 0.9472 (21%) 0.9243 (21%) 0.9778 (6%)
PLF (n) 0.9999 (0%) 0.9990 (0%) 0.9911 (5%) 0.9169 (16%) 0.7440 (47%)
PHF (n) 0.9982 (0%) 0.9645 (16%) 0.9316 (11%) 0.7970 (37%) 0.4859 (88%)
CLUSTER B
(n= 13)
RR (ip) 0.9828 (8%) 0.8996 (15%) 0.8690 (23%) 0.9258 (23%) 0.9473 (15%)
PLF (n) 0.9423 (8%) 0.8544 (23%) 0.9757 (8%) 0.8601 (15%) 0.8838 (15%)
PHF (n) 0.9284 (8%) 0.8710 (23%) 0.9112 (23%) 0.7455 (62%) 0.7402 (54%)
CLUSTER C
(n= 18)
RR (ip) 0.9943 (0%) 0.9363 (11%) 0.9665 (11%) 0.9035 (22%) 0.9615 (12%)
PLF (n) 0.9990 (0%) 0.9998 (0%) 0.9792 (6%) 0.9164 (28%) 0.7791 (31%)
PHF (n) 0.9909 (6%) 0.9670 (6%) 0.8906 (28%) 0.5914 (67%) 0.3843 (88%)
CLUSTER D
(n= 17)
RR (ip) 0.9945 (0%) 0.9884 (0%) 0.9844 (0%) 0.9751 (0%) 0.9541 (24%)
PLF (n) 0.9999 (0%) 0.9996 (0%) 0.9233 (18%) 0.9361 (18%) 0.9045 (24%)
PHF (n) 0.9929 (0%) 0.9739 (12%) 0.8100 (47%) 0.5381 (76%) 0.5366 (65%)
Values a e exp essed as CCC mean and (pe cen age o subjec s unde 0.9 h eshold). Segmen s a e based on he es phases: es ing
(S
REST
), cycling (S
CY
) and eco e y (S
REC
). S
CY
was di ided in h ee s ages a 60, 70 and 80% o HRmax, deno ed as S
CY60
, S
CY70
and S
CY80
espec i ely. The cha ac e is ics o each clus e we e he ollowing: CLUSTER A = high i ness; CLUSTER B = low i ness and low age;
CLUSTER C = low i ness, high age and medium unk a pe cen age; CLUSTER D = low i ness, high age and high unk a pe cen age.
RR (ip) = pai ed RR in e al se ies; PLF (n) = ins an aneous low- equency powe ; PHF (n) = ins an aneous high- equency powe .
Table 4shows CCC alues o RR (ip), P
LF
(n) and P
HF
(n) and ou lines he pe cen age
o subjec s no eaching excellen ag eemen (CCC > 0.90) o each segmen o he es . The
numbe o subjec s no eaching excellen ag eemen was clea ly highe o P
HF
(n) han o
Senso s 2021,21, 902 9 o 14
P
LF
(n) (
χ2
(deg ees o eedom);
χ2
(1) = 45.52; p< 0.001), i inc eased wi h exe cise in ensi y
(
χ2
(2) = 38.47; p< 0.001) and was lowe du ing exe cise han du ing S
REC
(
χ2
(1) = 42.31;
p< 0.001
). When pe o ming he analysis sepa a ely o each iden i ied clus e , CLUSTER
A ob ained he highes CCC alues, wi h CLUSTER D being he g oup wi h less subjec s
showing op imal ag eemen be ween de ices in P
HF
(n). Due o he p esence o noise in
he RR
P
(i) se ies du ing S
REC
, he ins an aneous powe could no be p ope ly ex ac ed in
4 olun ee s and he inal sample o SREC was N = 63.
Table 5shows %E o o each HRV index. K uskal-Wallis es demons a ed signi -
ican di e ences be ween clus e s in P
HF
a S
REST
and du ing exe cise (S
CY70
and S
CY80
).
Wi h ega ds o empo al domain HRV indices, SDNN showed signi ican di e ences
be ween g oups a S
REST
and du ing exe cise (S
CY60
and S
CY80
) and RMSSD showed signi -
ican di e ences a he highes in ensi ies (S
CY70
and S
CY80
). Bo h o P
HF
and o empo al
domain HRV indices, CLUSTER D was he g oup wi h he highes %E o . The magni ude
o all hese di e ences was small, i.e., <0.2 acco ding o he e ec sizes.
Table 5. Pe cen age o e o (%) o each HRV index and compa ison be ween clus e s.
CLUSTER A
(n= 19)
CLUSTER B
(n= 13)
CLUSTER C
(n= 18)
CLUSTER D
(n= 17)
Main E ec
pE ec Size
SREST
PLF 0.0 (−0.2 o 0.3) 0.2 (−0.3 o 0.6) 0.1 (−0.2 o 0.2) 0.3 (−0.1 o 0.4) 0.534 0.034
PHF −0.5 (−2.3 o 0.2) −0.2 (−0.7 o 0.4) −0.5 (−1.4 o 0.4) 0.5 (−0.3 o 2.0) 0.049 * 0.121
SDNN 0.0 (−0.2 o 0.5) B−0.1 (−0.8 o 0.0) A,D 0.0 (−0.2 o 0.2) 0.1 (0.0 o 0.2) B0.021 * 0.147
RMSSD −0.2 (−1.0 o 0.3) −0.1 (−2.3 o 0.4) −0.1 (−1.3 o 0.4) 0.2 (−0.2 o 2.3) 0.150 0.081
SCY60
PLF 0.8 (−0.7 o 18.7) 0.0 (−0.8 o 1.3) 0.4 (−4.0 o 3.0) 0.8 (−0.5 o 4.4) 0.444 0.041
PHF 25.7 (10.3 o 48.3) 2.1 (−12.5 o 17.7) 21.7 (−2.4 o 44.6) 27.0 (10.9 o 65.6) 0.164 0.077
SDNN 0.1 (−2.5 o 0.9) D−0.2 (−1.2 o 1.4) 1.2 (−0.1 o 4.2) 2.2 (1.1 o 8.5) A0.018 * 0.153
RMSSD 14.4 (4.6 o 32.4) 1.8 (−3.2 o 14.6) 21.7 (4.2 o 29.8) 15.8 (7.9 o 44.1) 0.266 0.060
SCY70
PLF 1.6 (−0.5 o 7.5) −0.2 (−2.8 o 0.9) −0.4 (−1.7 o 2.1) −0.6 (−7.8 o 2.8) 0.125 0.087
PHF 9.2 (−6.2 o 50.8) −0.9 (−42.6 o 25.6) C,D 25.3 (17.5 o 57.2) B51.8 (14.5 o 71.7) B0.007 * 0.186
SDNN 0.6 (−0.5 o 2.8) 0.2 (−1.4 o 1.6) 1.5 (−1.2 o 7.0) 2.3 (0.3 o 8.0) 0.104 0.093
RMSSD 15.4 (−2.0 o 29.5) 0.6 (−27.9 o 19.3) D15.6 (14.2 o 36.5) 27.1 (14.2 o 56.0) B0.010 * 0.172
SCY80
PLF 1.5 (−2.6 o 32.0) 2.8 (−1.0 o 3.6) 2.7 (−1.1 o 12.6) 5.9 (0.3 o 16.1) 0.596 0.029
PHF 31.6 (−27.4 o 87.7) −9.0 (−113.1 o 56.4) D28.0 (8.5 o 45.2) 44.7 (37.2 o 77.7) B0.047 * 0.121
SDNN 4.0 (−0.8 o 13.7) −0.4 (−2.9 o 6.1) 1.1 (−0.3 o 4.1) 5.7 (1.6 o 14.3) 0.050 * 0.119
RMSSD 19.2 (−9.9 o 67.0) 8.4 (−43.0 o 26.1) D19.0 (4.7 o 32.7) 32.9 (22.1 o 58.2) B0.028 * 0.137
SREC
PLF −0.4 (−33.8 o 0.1) 0.5 (−0.5 o 6.2) 0.2 (−17.6 o 3.9) 0.2 (−4.9 o 12.7) 0.161 0.105
PHF −1.6 (−8.2 o 5.8) 1.6 (0.1 o 5.9) −1.2 (−8.3 o 10.1) 9.3 (−0.2 o 37.2) 0.053 0.157
SDNN −0.1 (−0.9 o 0.1) 0.1 (−1.5 o 0.3) −0.1 (−1.0 o 0.4) 0.0 (−1.8 o 0.7) 0.933 0.007
RMSSD −0.9 (−2.3 o 2.0) 0.1 (−9.8 o 3.0) 1.5 (−1.4 o 5.5) 2.2 (−2.9 o 11.5) 0.226 0.068
Pe cen age e o (%) alues a e exp essed as median and in e qua ile ange. The cha ac e is ics o each clus e we e he ollowing:
CLUSTER A = high i ness; CLUSTER B = low i ness and low age; CLUSTER C = low i ness, high age and medium unk a pe cen age;
CLUSTER D = low i ness, high age and high unk a pe cen age. Segmen s a e based on he es phases: es ing (S
REST
), cycling
(S
CY
) and eco e y (S
REC
). S
CY
was di ided in h ee s ages a 60, 70 and 80% o HRmax, deno ed as S
CY60
, S
CY70
and S
CY80
espec i ely.
PLF = low- equency powe
; P
HF
= high- equency powe ; SDNN= SD o he RR in e als; RMSSD = oo mean squa e o successi e
di e ences be ween NN in e als. * = Signi ican di e ences be ween clus e s (p
≤
0.05, K uskal-Wallis es ).
A
= Di e en o CLUSTER A;
B= Di e en o CLUSTER B; C= Di e en o CLUSTER C; D= Di e en o CLUSTER D.
4. Discussion
In his s udy, HRV analysis om RR in e als p o ided by Pola H7 a es and du ing
a ious exe cise in ensi ies has been alida ed agains he same analysis om a simul a-
neous ECG eco ding. Wilcoxon es showed a la ge numbe o signi ican di e ences
be ween de ices in HRV indices du ing exe cise. Howe e , he e ec size was small o
medium and o li le p ac ical ele ance in he case o MHR. When obse ing RR (ip),
P
LF
(n) and P
HF
(n) signals, he pe cen age o subjec s no eaching excellen ag eemen
be ween de ices (CCC > 0.90) inc eased wi h exe cise in ensi y and was highe o P
HF
(n)
han o P
LF
(n). Clus e analysis e ealed ha pheno ypic cha ac e is ics like age, body
composi ion and i ness le el in luenced HRV measu emen s as well as he di e ences
be ween Pola H7 and ECG. In pa icula , CLUSTER D, composed o subjec s wi h low
i ness le el, high age and high unk a pe cen age, was he g oup wi h he lowes numbe