T
E
L
KO
M
NI
K
A
,
V
ol
.
14,
N
o.
3,
S
ept
em
ber
20
16,
pp.
99
9~
1
008
I
S
S
N
:
1
693
-
6
930
,
ac
c
r
edi
t
ed
A
b
y
D
IK
T
I,
D
e
c
r
e
e
N
o
:
58/
D
I
K
T
I
/
K
ep/
2013
D
O
I
:
10.
12928/
T
E
LK
O
M
N
I
K
A
.
v
1
4
i
3
.
3665
99
9
R
ec
ei
v
ed
A
p
r
il
2
,
2
01
6
;
R
e
v
i
s
ed
J
une
1
0
,
20
1
6
;
A
c
c
e
pt
ed
J
u
ne
2
9
,
201
6
V
e
ntr
i
c
ul
a
r
Ta
c
h
y
a
r
r
h
y
t
h
mi
a
O
ns
e
t P
r
e
di
c
ti
on
B
ased
on H
R
V
a
n
d G
e
ne
ti
c
A
l
g
or
i
th
m
K.
H.
Bo
o
n
*
1
,
M
B
M
a
la
r
v
ili
2
, M
. K
h
a
lil
-
H
an
i
3
F
ac
u
l
t
y
of
E
l
ec
t
r
i
c
a
l
E
ng
i
ne
er
i
ng,
U
ni
v
er
s
i
t
i
T
ek
ono
l
og
i
M
al
ay
s
i
a,
81310
,
S
k
udai
,
J
ohor
,
M
al
ay
s
i
a
.
*
C
or
r
es
po
ndi
ng a
ut
hor
,
e
-
m
ai
l
:
boon.
k
h
ang.
hua@
g
m
ai
l
.
c
om
1
,
m
a
la
r
v
i
l
i
@
bi
om
e
d
ic
al
.
u
t
m
.
m
y
2
,
k
hal
i
l
@
f
k
e.
ut
m
.
m
y
3
A
b
st
r
act
P
r
edi
c
t
i
ng
on
s
et
of
v
e
nt
r
i
c
ul
ar
t
ac
h
y
ar
r
h
y
t
hm
i
a
pr
o
v
i
des
op
p
or
t
uni
t
i
e
s
t
o r
ed
uc
e c
as
ual
t
i
es
due
t
o
s
udd
en
c
ar
di
a
c
d
eat
h
.
H
ow
e
v
er
,
t
h
e pr
edi
c
t
i
on
ac
c
ur
a
c
y
s
t
i
l
l
ne
e
ds
i
m
pr
ov
em
ent
.
T
her
e
f
or
e,
w
e
ai
m
t
o
pr
opo
s
e
a
m
et
h
od
t
h
at
c
a
n
pr
edi
c
t
t
he
o
ns
et
of
t
a
c
h
y
ar
r
h
y
t
h
m
i
a
ev
en
t
s
w
i
t
h
i
m
pr
ov
ed
a
c
c
ur
ac
y
ba
s
ed
o
n
hear
t
r
a
t
e v
ar
i
a
bi
l
i
t
y
and S
u
p
por
t
V
e
c
t
or
M
ac
hi
n
e c
l
as
s
i
f
i
er
.
F
i
f
t
y
per
c
ent
of
s
am
pl
e
dat
a
f
r
om
s
t
an
dar
d
dat
ab
as
e
w
as
us
ed
t
o t
r
ai
n
t
he c
l
a
s
s
i
f
i
er
,
an
d t
h
e r
em
ai
n
der
w
a
s
u
s
ed
t
o
v
er
i
f
y
t
he
p
er
f
or
m
anc
e.
F
i
v
e
mi
n
u
t
es
R
R
i
n
t
er
v
al
s
i
m
m
edi
a
t
el
y
pr
i
or
t
o
t
a
c
h
y
ar
r
h
y
t
hm
i
a
e
v
ent
f
r
om
e
ac
h
s
am
pl
e
dat
a
w
as
c
r
opp
ed
f
or
ec
t
o
pi
c
be
at
c
or
r
ec
t
i
on an
d t
h
en c
on
v
er
t
e
d t
o hear
t
r
at
e.
E
x
t
r
ac
t
i
o
n of
t
i
m
e dom
ai
n,
s
p
e
c
t
r
al
,
n
on
-
l
i
near
and
b
i
s
p
ec
t
r
um
f
eat
ur
es
w
er
e
per
f
or
m
e
d
s
u
bs
e
que
nt
l
y
.
F
ur
t
her
m
or
e,
gen
et
i
c
al
gor
i
t
h
m
w
as
us
e
d
t
o
s
i
m
ul
t
a
neo
us
l
y
o
pt
i
m
i
z
e t
he f
eat
ur
e
s
u
bs
e
t
an
d c
l
a
s
s
i
f
i
er
par
am
et
er
s
.
W
i
t
h t
he
opt
i
m
i
z
at
i
on
,
pr
e
di
c
t
i
o
n
ac
c
ur
ac
y
of
our
pr
opo
s
ed
m
et
hod
a
bl
e t
o ou
t
per
f
or
m
pr
ev
i
ou
s
w
or
k
s
w
i
t
h
77.
94%,
80.
8
8%
and 7
9.
41
% f
or
se
n
si
t
i
v
i
t
y
,
s
pe
c
i
f
i
c
i
t
y
and ac
c
u
r
ac
y
r
es
pec
t
i
v
el
y
.
K
eyw
o
r
d
s
:
H
ear
t
R
at
e
V
ar
i
abi
l
i
t
y
,
A
r
r
h
y
t
hm
i
a P
r
ed
i
c
t
i
on,
V
ent
r
i
c
u
l
ar
T
ac
h
y
ar
r
h
y
t
hm
i
a
(
V
T
A
)
,
G
e
n
et
i
c
A
l
gor
i
t
hm
,
B
i
s
pec
t
r
um
f
eat
ur
e
s
.
C
o
p
y
r
i
g
h
t
©
20
16 U
n
i
ver
si
t
a
s A
h
mad
D
ah
l
an
.
A
l
l
r
i
g
h
t
s r
eser
ved
.
1.
I
n
tr
o
d
u
c
ti
o
n
V
ent
r
i
c
ul
ar
t
ac
h
y
ar
r
h
y
t
hm
i
a
(
V
T
A
)
i
s
a
t
y
pe
of
t
he
ar
r
hy
t
hm
i
a
(
abnor
m
al
hear
t
r
h
y
t
hm
s
)
t
hat
ar
i
s
es
f
r
o
m
i
m
pr
oper
el
ec
t
r
i
c
al
ac
t
i
v
i
t
y
i
n t
he b
ot
t
o
m
c
ha
m
ber
s
of
t
he
hear
t
c
al
l
e
d v
ent
r
i
c
l
es
.
O
c
c
ur
r
enc
e
of
V
T
A
i
s
h
a
r
m
f
ul
t
o
pat
i
en
t
’
s
h
ea
l
t
h
bec
aus
e
i
t
c
a
us
es
f
ai
nt
i
n
g,
p
al
p
i
t
a
t
i
o
ns
,
as
y
s
t
ol
e,
an
d
ev
e
n
t
r
i
g
ger
s
appr
ox
i
m
at
el
y
8
0%
of
s
udden
c
ar
di
ac
deat
h
(
S
C
D
)
ca
se
s
[
1]
.
S
CD
ac
c
ount
s
f
or
one o
ut
of
ev
er
y
t
w
o de
at
hs
f
r
om
c
ar
di
o
v
as
c
ul
ar
di
s
e
as
es
[
2]
.
T
her
ef
or
e,
dev
el
opm
ent
of
t
he r
el
i
a
bl
e
pr
edi
c
t
or
of
V
T
A
ons
et
,
na
m
el
y
t
h
e v
e
nt
r
i
c
u
l
ar
t
ac
h
y
c
ar
di
a (
V
T
)
an
d
v
en
t
r
i
c
ul
ar
f
i
br
i
l
l
at
i
o
n (
V
F
)
,
i
s
c
l
i
ni
c
a
l
i
m
por
t
ant
bec
a
u
s
e t
he
pr
ed
i
c
t
i
o
n pr
o
v
i
des
oppor
t
un
i
t
y
t
o
t
i
m
el
y
pr
ev
ent
i
on of
t
he ne
gat
i
v
e c
ons
e
que
nc
es
br
ou
ght
b
y
V
T
A
t
hr
oug
h ear
l
y
t
er
m
i
nat
i
on b
y
i
m
pl
ant
ab
l
e c
ar
d
i
o
v
er
t
er
de
f
i
br
i
l
l
at
or
(
I
C
D
)
[
3]
.
R
R
i
nt
er
v
a
l
i
s
t
he
t
i
m
e
i
nt
e
r
v
al
bet
w
een
t
w
o
c
ons
ec
ut
i
v
e
h
ear
t
b
eat
s
w
h
i
l
e
he
ar
t
r
at
e
i
s
t
he r
ec
i
pr
oc
a
l
of
R
R
i
n
t
er
v
a
l
.
H
ear
t
r
a
t
e
v
ar
i
ab
i
l
i
t
y
(
H
R
V
)
s
i
g
nal
i
s
v
ar
i
at
i
on
of
he
a
r
t
r
at
e t
hat
c
an
be
us
ed
t
o
d
i
ag
nos
e
c
ar
d
i
ov
as
c
u
l
ar
di
s
e
as
es
.
G
en
er
al
l
y
,
H
R
V
s
i
gna
l
c
an
b
e
ob
t
ai
n
ed
t
hr
oug
h
m
eas
ur
i
ng t
he
c
ons
ec
ut
i
v
e
R
p
eak
s
of
el
ec
t
r
oc
ar
di
o
gr
am
(
E
C
G
)
s
i
gnal
[
4,
5]
.
A
not
her
m
or
e
c
onv
e
ni
ent
m
et
hod
t
o
obt
a
i
n
H
R
V
i
s
us
i
ng
t
he
t
i
m
e
di
f
f
er
enc
e
bet
w
een
t
w
o
c
o
ns
ec
ut
i
v
e
pul
s
es
i
n t
he p
hot
o
pl
et
h
y
s
m
ogr
ap
h (
P
P
G
)
s
i
gna
l
[
6]
.
H
R
V
s
i
gna
l
i
s
us
ual
l
y
a
nal
y
z
ed
w
i
t
h v
ar
i
ous
H
R
V
ana
l
y
s
i
s
bas
e
d f
eat
ur
e ex
t
r
ac
t
i
on t
ec
h
ni
ques
.
T
hes
e t
ec
hni
ques
ha
v
e bee
n w
i
d
el
y
app
l
i
ed i
n
v
ar
i
ous
m
edi
c
al
r
el
at
e
d r
e
s
ear
c
hes
s
uc
h as
c
l
as
s
i
f
i
c
at
i
o
n of
c
ar
di
a ar
r
h
y
t
hm
i
a,
di
a
gnos
i
s
of
neon
at
a
l
s
eps
i
s
,
di
s
c
r
i
m
i
na
t
i
on
of
s
l
eep
s
t
age
an
d et
c
.
,
[
7]
.
H
R
V
ana
l
y
s
i
s
i
s
a
l
s
o
o
ne
of
t
he
p
opu
l
ar
m
et
hods
t
h
at
i
s
ac
t
i
v
el
y
be
i
ng
r
es
ear
c
hed
f
or
app
l
i
c
at
i
on
i
n
V
T
A
o
ns
et
pr
ed
i
c
t
i
o
n.
I
n
i
t
i
al
l
y
,
pr
e
v
i
ous
w
or
k
s
hav
e
f
oc
us
ed
t
hei
r
s
t
ud
y
on
s
t
at
i
s
t
i
c
a
l
di
f
f
er
enc
e of
t
he H
R
V
f
eat
ur
es
(
f
eat
ur
es
ex
t
r
ac
t
ed bas
ed o
n di
f
f
er
e
nt
H
R
V
ana
l
y
s
i
s
t
ec
hni
ques
)
t
hat
ex
t
r
ac
t
ed
f
r
o
m
t
he hear
t
r
at
e pr
i
or
t
o V
T
A
o
ns
et
a
nd t
he c
o
nt
r
ol
d
at
a (
H
R
V
s
i
gna
l
w
i
t
h
out
V
T
A
e
v
e
nt
s
)
r
es
pec
t
i
v
el
y
.
Ma
n
y
r
ep
or
t
s
ha
v
e f
oun
d t
hat
t
her
e
ar
e s
t
at
i
s
t
i
c
al
s
i
gni
f
i
c
an
t
c
hang
es
i
n H
R
V
f
eat
ur
es
v
al
ues
pr
i
or
t
o
V
T
A
s
[8
-
10]
.
I
ns
pi
r
ed b
y
t
hes
e f
i
ndi
ngs
,
v
ar
i
ous
V
T
A
o
ns
et
pr
edi
c
t
i
o
n m
et
hods
bas
ed
on
H
R
V
hav
e
b
een de
v
e
l
op
ed.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
99
–
1
0
08
1000
H
ow
ev
e
r
,
t
he m
ai
n
i
s
s
ue i
s
t
hat
t
he
pr
ed
i
c
t
i
o
n r
es
ul
t
s
of
t
he pr
e
v
i
ous
w
or
k
s
ar
e s
t
i
l
l
uns
at
i
s
f
ac
t
or
y
f
or
pr
edi
c
t
i
n
g s
uc
h l
i
f
e
-
t
hr
eat
e
ni
ng ar
r
h
y
t
hm
i
a.
A
m
ong t
he pr
ev
i
ous
w
or
k
s
,
t
he
hi
g
hes
t
ac
hi
e
v
a
bl
e pr
e
di
c
t
i
on ac
c
ur
ac
y
w
as
75.
6%
[
1
1]
.
O
t
h
er
pr
e
v
i
ous
w
or
k
s
c
o
ul
d
not
ac
h
i
e
v
e
t
hat
ac
c
ur
ac
y
l
e
v
e
l
.
B
y
u
s
i
ng h
ear
t
r
at
e
pat
t
er
n,
T
hong
and
R
ai
t
t
[
1
2]
ac
h
i
e
v
ed pr
ed
i
c
t
i
o
n
per
f
or
m
anc
e
w
i
t
h 53%
of
s
ens
i
t
i
v
i
t
y
and 91%
of
s
pec
i
f
i
c
i
t
y
.
W
i
t
h
dec
i
s
i
o
n r
ul
e
bas
ed s
y
s
t
em
bas
ed
on
m
ul
t
i
po
l
e
an
al
y
s
i
s
,
R
o
z
en
,
et
a
l
.
,
[
13]
ac
hi
e
v
e
d 5
0%
of
s
ens
i
t
i
v
i
t
y
and
91.
6
%
of
s
p
e
c
if
ic
it
y
.
W
o
llm
a
n
,
et
al
.
,
[
14]
pr
o
pos
ed
a
m
et
hod,
w
hi
c
h
p
er
f
or
m
ed
t
he
V
T
A
ons
et
pr
e
di
c
t
i
o
n
bas
ed on t
i
m
e dom
ai
n f
ea
t
ur
es
and r
egr
es
s
i
on t
r
e
e c
l
as
s
i
f
i
er
s
,
obt
ai
ned 70
.
9%
of
ac
c
ur
ac
y
.
J
oo
,
et
al
.
,
[
11]
pr
opos
e
d a pr
edi
c
t
i
on m
et
hod t
h
at
at
t
a
i
ne
d t
he m
os
t
bal
anc
e
d and h
i
gh
es
t
pr
edi
c
t
i
on p
er
f
or
m
anc
e a
m
ong pr
e
v
i
ous
w
or
k
s
w
i
t
h 77
.
3%
of
s
ens
i
t
i
v
i
t
y
,
73.
8%
of
s
pec
i
f
i
c
i
t
y
a
nd
75.
6%
of
ac
c
ur
ac
y
.
F
ur
t
her
m
or
e,
t
hei
r
m
et
hod
us
ed
s
i
gni
f
i
c
ant
l
y
s
hor
t
er
H
R
V
s
i
g
nal
(
5
m
i
nut
es
onl
y
)
t
o ac
hi
ev
e hi
g
her
pr
edi
c
t
i
on ac
c
ur
ac
y
.
I
n c
ont
r
as
t
,
ot
her
pr
e
v
i
o
us
w
or
k
s
[1
2
-
14]
u
t
i
liz
e
d
m
or
e t
han 10
m
i
nut
es
of
H
R
V
s
i
gn
al
i
n
l
e
ngt
h pr
i
or
t
o
V
T
A
e
v
ent
f
or
pr
ed
i
c
t
i
on.
T
he m
ai
n
r
eas
ons
f
or
s
uc
h h
i
g
her
pr
e
di
c
t
i
on
per
f
or
m
anc
e
w
er
e
u
t
i
l
i
z
at
i
on
of
m
or
e c
o
m
pr
ehens
i
v
e t
y
pes
of
H
R
V
f
eat
ur
es
t
han
ot
her
pr
e
v
i
o
us
w
or
k
s
and
em
pl
o
y
m
ent
of
m
or
e
adv
anc
ed
s
u
per
v
i
s
ed
c
la
s
s
if
ie
r
-
ar
t
i
f
i
c
i
a
l
n
eur
a
l
net
w
or
k
(
A
N
N
)
.
I
n
t
he
i
r
w
or
k
,
m
ul
t
i
pl
e c
a
t
eg
or
i
es
of
H
R
V
f
eat
ur
es
bas
ed on t
i
m
e do
m
ai
n an
al
y
s
i
s
,
s
pec
t
r
al
an
al
y
s
i
s
ba
s
ed on f
as
t
F
ou
r
i
er
t
r
ans
f
or
m
(
F
F
T
)
,
and
poi
nc
ar
e p
l
ot
w
er
e
em
pl
o
y
e
d t
o t
r
a
i
n
t
he m
odel
of
t
he
A
N
N
.
T
o
addr
es
s
t
he
af
or
em
ent
i
oned
i
s
s
ue,
t
he
m
ai
n
obj
ec
t
i
v
e
of
t
h
i
s
pa
per
i
s
t
o
pr
op
os
e
a
V
T
A
ons
et
pr
ed
i
c
t
i
on
m
et
hod
w
i
t
h
i
m
pr
ov
ed
pr
ed
i
c
t
i
o
n
ac
c
ur
ac
y
.
C
on
t
r
i
b
ut
i
ons
of
our
w
or
k
c
an
be
s
um
m
ar
i
z
ed
as
f
ol
l
o
w
s
.
I
ns
pi
r
ed
b
y
J
o
o
,
et
al
.
,
[
11]
,
t
hi
s
pa
per
a
l
s
o
em
pl
o
y
e
d
m
ul
t
i
pl
e
t
y
p
es
of
H
R
V
f
eat
ur
es
f
or
pr
edi
c
t
i
on
.
H
o
w
e
v
er
,
a
dd
i
t
i
ona
l
t
y
p
e of
H
R
V
f
eat
ur
es
s
uc
h
as
T
r
i
angu
l
ar
I
nt
er
po
l
at
i
o
n of
R
R
i
nt
er
v
a
l
hi
s
t
ogr
am
,
s
a
m
pl
e ent
r
o
p
y
and h
i
gh
er
or
der
s
pec
t
r
al
a
nal
y
s
i
s
,
w
h
i
c
h
hav
e
not
bee
n
us
ed
i
n
[
1
1]
,
ar
e
al
s
o
us
ed
t
o
t
r
a
i
n
S
u
ppor
t
v
ec
t
or
m
ac
hi
ne
(
S
V
M)
c
l
as
s
i
f
i
er
f
or
pr
edi
c
t
i
on i
n our
w
or
k
.
F
ur
t
her
m
or
e,
t
he genet
i
c
al
gor
i
t
hm
(
G
A
)
bas
ed f
eat
ur
e s
el
ec
t
i
on pr
oc
es
s
pr
opos
e
d b
y
H
uan
g a
nd
W
a
ng
[
15]
is
a
dopt
ed
t
o
s
i
m
ul
t
aneous
l
y
opt
i
m
i
z
e t
h
e H
R
V
f
eat
ur
e
s
ubs
et
an
d
S
V
M c
l
as
s
i
f
i
er
par
am
et
er
s
.
W
i
t
h t
he a
ddi
t
i
ona
l
t
y
pes
of
H
R
V
f
eat
ur
es
and
G
A
b
as
e
d
opt
i
m
i
z
at
i
o
n pr
oc
es
s
,
pr
ed
i
c
t
i
on
per
f
or
m
anc
e of
our
m
et
ho
d o
ut
per
f
or
m
s
al
l
pr
e
v
i
ous
w
or
k
s
w
i
t
h
77.
9
4%
,
80.
88%
an
d
79.
41
%
f
or
s
ens
i
t
i
v
i
t
y
,
s
pec
i
f
i
c
i
t
y
and
ac
c
ur
ac
y
r
es
p
ec
t
i
v
el
y
ev
en
t
h
oug
h
w
e us
e s
t
r
i
c
t
er
appr
o
ac
h t
o ev
al
u
at
e o
ur
m
et
hod
.
P
r
edi
c
t
i
on
s
ens
i
t
i
v
i
t
y
a
nd s
p
ec
i
f
i
c
i
t
y
of
our
m
et
hod
ar
e
m
or
e
bal
anc
e
d
w
h
en
c
om
par
ed
t
o
pr
ev
i
o
u
s
w
or
k
s
[
12
-
14]
.
F
ur
t
her
m
o
r
e,
i
n
c
ont
r
as
t
t
o
m
os
t
o
f
pr
ev
i
ous
w
or
k
s
[
12
-
14]
t
hat
us
e
d
m
or
e
t
han
10
m
i
nut
es
of
H
R
V
s
i
gn
al
,
our
m
et
hod
onl
y
us
es
5
m
i
nut
es
H
R
V
s
i
gn
al
,
wh
i
c
h
en
d
i
m
m
edi
at
e
l
y
pr
i
or
t
o
V
T
A
ons
et
,
f
or
f
eat
ur
e
ex
t
r
ac
t
i
on.
F
i
na
l
l
y
,
opt
i
m
al
H
R
V
f
eat
ur
e s
u
bs
et
s
el
ec
t
e
d b
y
G
A
f
or
pr
edi
c
t
i
on i
s
a
l
s
o r
epor
t
ed f
or
r
ef
er
enc
e i
n f
ut
ur
e
w
or
k
s
.
T
he l
a
y
o
ut
of
t
he
pap
er
i
s
as
f
ol
l
o
w
s
.
S
ec
t
i
o
n 2
pr
es
ent
s
t
h
e da
t
ab
as
e a
nd
pr
o
pos
ed
m
et
hod.
S
ec
t
i
on
3 pr
es
e
nt
s
t
he r
es
u
l
t
a
nd
ana
l
y
s
i
s
.
S
e
c
t
i
on
4 pr
es
e
nt
s
t
he
c
onc
l
u
s
i
on.
2.
P
r
o
p
o
s
e
d
V
T
A
o
n
s
e
t P
r
e
d
i
c
ti
o
n
M
e
th
o
d
B
lo
c
k
di
agr
am
i
n
F
i
gur
e
1
s
ho
w
s
t
he
ov
er
v
i
e
w
of
t
he
pr
opos
e
d
m
et
hod.
I
t
c
o
m
pr
i
s
es
of
pr
e
-
pr
oc
es
s
i
ng s
t
ag
e,
H
R
V
f
eat
ur
e ex
t
r
ac
t
i
on
s
t
age,
H
R
V
f
eat
ur
e s
el
ec
t
i
on
s
t
ag
e and s
uppor
t
v
ec
t
or
m
ac
hi
ne (
S
V
M)
bas
ed c
l
as
s
i
f
i
c
at
i
on s
t
age.
Pre
-
Processing
HRV Feature
Extraction
SVM
5
minutes RR
Intervals
HRV Features
GA
Feature
Selection
VTA Onset
Prediction
Optimization of Feature Set and Classifier
F
i
gur
e
1.
O
v
er
v
i
e
w
of
P
r
op
os
ed M
et
ho
d
F
i
r
s
t
l
y
,
5
m
i
nut
es
R
R
i
nt
er
v
al
s
t
h
at
e
nd
i
m
m
edi
at
el
y
pr
i
or
t
o
V
T
A
ons
et
f
or
106
pr
e
-
VT
/
VF
d
a
t
a
and
5 m
i
nut
es
R
R
i
nt
er
v
a
l
s
of
c
ont
r
ol
da
t
a
f
r
om
t
he dat
a
bas
e
ar
e f
ed t
o t
he
pr
e
-
pr
oc
es
s
i
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
V
ent
r
i
c
ul
ar
T
ac
hy
ar
r
hy
t
hm
i
a O
ns
et
P
r
e
d
ic
t
io
n
B
a
s
ed
o
n H
R
V
an
d G
en
et
i
c
…
(
K.
H
.
Bo
o
n
)
1001
s
t
age r
es
p
ec
t
i
v
e
l
y
.
Pr
e
-
pr
oc
es
s
i
ng s
t
age
i
nc
l
u
des
e
c
t
opi
c
beat
c
or
r
ec
t
i
o
n,
r
es
am
pl
i
ng
of
R
R
i
nt
er
v
a
l
a
nd
c
on
v
er
s
i
on
t
o
H
R
V
s
e
que
nc
es
b
y
c
om
put
i
ng
t
he
r
ec
i
pr
oc
a
l
of
R
R
i
n
t
er
v
a
l
.
I
n
H
R
V
f
eat
ur
e ex
t
r
ac
t
i
on s
t
a
ge,
5
t
im
e
-
dom
ai
n,
6 f
r
equenc
y
-
dom
ai
n,
5 n
on
-
l
i
n
ear
a
nd
37 b
i
s
pec
t
r
um
f
eat
ur
es
ar
e
t
h
en
ex
t
r
ac
t
e
d f
r
o
m
quant
i
f
i
e
d H
R
V
.
D
ur
i
ng
t
he f
eat
u
r
e s
e
l
ec
t
i
on
s
t
age
,
gen
et
i
c
al
g
or
i
t
hm
(
G
A
)
i
s
us
ed
t
o
opt
i
m
i
z
e t
he
f
eat
ur
e
s
et
.
F
i
n
al
l
y
,
pr
e
di
c
t
i
on
per
f
or
m
anc
e of
t
he
pr
opos
e
d m
et
hod i
s
ev
al
u
at
ed
i
n t
er
m
of
s
ens
i
t
i
v
i
t
y
,
s
pec
i
f
i
c
i
t
y
a
nd ac
c
ur
ac
y
.
D
e
t
a
il o
f
e
ac
h
bl
oc
k
i
s
des
c
r
i
bed
i
n
t
he
r
e
m
ai
ni
ng
s
ub
-
s
ec
t
i
ons
.
2.
1.
D
at
ab
as
e
R
R
i
nt
er
v
a
l
s
r
ec
or
di
ng ar
e obt
a
i
ne
d f
r
o
m
t
he S
p
ont
a
n
eous
V
ent
r
i
c
u
l
ar
T
ac
h
y
ar
r
h
y
t
hm
i
a
D
at
ab
as
e V
er
s
i
on
1.
0 f
r
om
Medt
r
o
ni
c
,
I
nc
.
f
r
o
m
t
he P
h
y
s
i
o
net
[
16]
.
T
he d
at
ab
as
e
w
as
c
ol
l
ec
t
ed
f
r
o
m
r
ec
or
ds
of
78
pat
i
ent
s
w
i
t
h
I
C
D
s
(
63
m
al
es
and
15
f
em
al
es
,
aged
f
r
om
20.
7
t
o
7
5.
3)
a
nd
c
ons
i
s
t
ed of
t
he f
ol
l
o
w
i
ng
R
R
i
nt
er
v
a
l
s
:
106 pr
e
-
V
T
r
ec
or
ds
,
29 pr
e
-
V
F
r
ec
or
ds
,
and 135 c
ont
r
o
l
dat
a
s
et
s
.
E
ac
h
d
at
a
i
nc
l
u
ded
102
4
R
R
i
nt
er
v
al
s
(
c
or
r
es
pond
i
ng
t
o
ar
ou
nd
15
m
i
nut
es
)
.
S
hor
t
-
t
er
m
H
R
V
ana
l
y
s
i
s
i
s
per
f
or
m
ed on t
he 5
m
i
nut
es
R
R
i
nt
er
v
al
s
pr
i
or
t
o
e
a
c
h
VT
A(
VT
/
VF
)
ev
e
nt
.
2.
2.
P
r
ep
r
o
ce
ssi
n
g
R
R
i
nt
er
v
a
l
s
ar
e
ob
t
ai
ne
d f
r
o
m
t
he dat
abas
e.
H
R
V
i
s
t
hen f
or
m
ed b
y
c
om
put
i
ng t
he
r
ec
i
pr
oc
al
of
t
he
i
nt
er
v
a
l
s
b
et
w
ee
n
s
uc
c
es
s
i
v
e
R
p
eak
s
.
A
f
t
er
t
hat
,
t
he
s
i
gn
al
i
s
r
es
am
pl
ed
t
o
4
H
z
s
i
gn
al
b
y
us
i
n
g
t
he c
ub
i
c
s
pl
i
n
e i
nt
er
pol
at
i
on
as
i
t
has
been r
ep
or
t
ed
t
h
at
t
h
i
s
t
ec
hni
qu
e
i
s
bet
t
er
t
han
l
i
ne
ar
i
n
t
er
po
l
at
i
on
[
1
7]
.
B
ef
or
e t
h
e H
R
V
i
s
r
es
am
pl
e
d,
i
t
is
e
v
al
u
at
e
d an
d
c
or
r
ec
t
ed bas
e
d o
n Mc
N
a
m
e
s
’
s
al
g
or
i
t
hm
[
18]
.
T
hi
s
al
gor
i
t
h
m
det
ec
t
s
t
he a
bnor
m
al
he
ar
t
r
at
e s
i
g
na
l
i
n t
he s
er
i
es
w
hi
c
h m
a
y
be
c
aus
ed b
y
ec
t
op
i
c
be
at
,
ar
t
i
f
ac
t
noi
s
e or
m
i
s
s
peak
det
ec
t
i
on
.
T
he al
gor
i
t
hm
ev
al
u
at
es
he
ar
t
r
at
e
us
i
ng
t
es
t
s
t
at
i
s
t
i
c
,
D
(
n)
c
a
l
c
ul
at
e
d
w
i
t
h f
ol
l
o
w
i
n
g eq
uat
i
on:
(
)
=
|
(
)
−
|
1
.
4
8
3
{
|
(
)
−
|
}
(
1)
W
h
er
e H
R
m
i
s
t
he av
er
age
hear
t
r
at
e
of
t
he s
er
i
es
,
H
R
(
n)
i
s
i
ns
t
ant
ane
ous
he
ar
t
r
at
e an
d m
ed
{
}
i
s
m
edi
an
f
i
l
t
er
.
I
f
D
(
n)
v
al
u
e
ex
c
es
s
c
er
t
ai
n
t
hr
es
hol
d,
τ
,
t
he
i
ns
t
ant
ane
ous
hear
t
r
at
e
i
s
c
ons
i
der
e
d as
a
bnor
m
al
he
ar
t
r
at
e.
T
he h
ear
t
r
at
e i
s
c
or
r
ec
t
ed
w
i
t
h f
ol
l
o
w
i
n
g eq
u
at
i
o
n:
(
)
=
(
+
)
:
|
|
<
−
1
2
(
2)
W
h
er
e
w
m
i
s
t
he
w
i
n
do
w
l
e
ngt
h
of
t
he
m
edi
um
f
i
l
t
er
.
I
n
t
h
i
s
w
or
k
,
τ
an
d
w
m
ar
e
s
e
t
to
4
an
d
1
1
r
es
pec
t
i
v
el
y
ac
c
or
di
n
g t
o t
h
e l
i
t
er
at
ur
e r
e
v
i
e
w
[
1
8]
.
F
i
gur
e
2 s
ho
w
s
5 m
i
nut
es
R
R
i
nt
er
v
al
s
of
a pat
i
e
nt
pr
i
or
t
o V
T
A
ev
ent
.
R
R
i
nt
er
v
a
l
s
bef
or
e an
d af
t
er
c
or
r
ec
t
i
on
b
y
Mc
N
am
es
’
s
al
gor
i
t
hm
ar
e s
ho
w
n
.
F
i
gur
e
2.
F
i
v
e
m
i
nut
es
R
R
i
nt
er
v
al
s
pr
i
or
t
o
V
T
A
e
v
en
t
bef
or
e an
d af
t
er
ec
t
op
i
c
be
at
c
or
r
ec
t
i
o
n
0
1
2
3
4
5
0.
7
0.
72
0.
74
0.
76
0.
78
T
i
m
es
(
m
i
nut
e)
R
R
I
nt
er
v
al
(
s
ec
ond)
R
R
i
nt
er
v
al
s
pr
i
or
t
o V
T
A
ev
ent
s
(
af
t
er
c
or
r
ec
t
i
on)
0
1
2
3
4
5
0.
4
0.
6
0.
8
1
T
i
m
es
(
m
i
nut
e)
R
R
I
nt
er
v
al
(
s
ec
ond)
R
R
i
nt
er
v
al
s
pr
i
or
t
o V
T
A
ev
ent
s
(
bef
or
e c
or
r
ec
t
i
on)
a
b
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
99
–
1
0
08
1002
2.
3.
H
R
V
f
e
at
u
r
e
ext
r
act
i
o
n
I
n t
h
i
s
s
t
ud
y
,
a n
um
ber
of
53 H
R
V
f
eat
ur
es
ar
e
ex
t
r
ac
t
ed f
r
om
H
R
V
us
i
n
g t
i
m
e
-
d
om
ai
n,
f
r
equenc
y
dom
ai
n
a
nd
non
-
l
i
ne
ar
a
nal
y
s
i
s
.
E
ac
h
of
f
eat
ur
es
a
nd
i
t
s
a
bbr
e
v
i
at
i
on
s
ar
e ex
p
l
ai
ne
d
i
n t
h
e f
ol
l
o
w
i
ng s
ub
-
s
ec
t
i
on
s
.
A
l
l
of
t
he m
ent
i
on
ed f
eat
ur
es
ar
e
w
e
l
l
k
no
w
n
a
nd h
av
e
bee
n us
ed
i
n o
t
her
r
e
l
at
ed H
R
V
s
t
u
di
e
s
[
19]
.
2.
3.
1
.
T
i
m
e
D
o
m
a
i
n
F
e
a
tu
r
e
s
S
ix
t
im
e
-
dom
ai
n
H
R
V
f
eat
ur
es
ar
e
c
o
m
put
ed
b
y
us
i
ng
s
t
at
i
s
t
i
c
al
a
nal
y
s
i
s
.
T
he
y
a
r
e
t
he
m
ean
of
H
R
V
(
MeanR
R
)
,
t
he
s
t
and
ar
d
de
v
i
at
i
on
of
H
R
V
(
S
D
R
R
)
,
r
oot
m
ean
s
qu
ar
e
of
s
uc
c
es
s
i
v
e d
i
f
f
er
enc
e i
nt
er
v
a
l
s
(
R
MS
S
D
)
,
num
ber
of
adj
ac
ent
R
R
i
nt
er
v
a
l
s
di
f
f
er
i
ng
b
y
m
or
e
t
han
50
m
s
(
N
N
50)
,
and
s
um
o
f
N
N
50
d
i
v
i
d
ed
b
y
t
he
t
ot
a
l
num
ber
of
al
l
R
R
i
nt
e
r
v
al
s
(
pN
N
50)
.
B
es
i
d
es
,
H
R
V
t
r
i
ang
ul
ar
i
n
dex
(
H
R
V
T
r
i
)
[
20]
w
as
a
l
s
o em
pl
o
y
e
d t
o ex
t
r
ac
t
g
eo
m
et
r
i
c
al
H
R
V
f
eat
ur
e.
H
R
V
T
r
i
i
s
t
he
t
o
t
al
num
ber
N
N
i
nt
er
v
a
l
s
di
v
i
d
ed
b
y
num
ber
of
R
R
i
nt
er
v
al
s
t
hat
f
al
l
t
o
m
odal
bi
n
.
2.
3.
2
S
p
e
ct
r
al
F
eat
u
r
es
F
or
s
pec
t
r
al
ana
l
y
s
i
s
,
po
w
er
s
pec
t
r
al
dens
i
t
y
(
P
S
D
)
w
as
es
t
i
m
at
ed
f
r
o
m
H
R
V
s
i
gn
al
.
I
t
i
s
ge
ner
a
l
l
y
ac
c
ept
e
d t
hat
s
pec
t
r
al
po
w
er
i
n
l
o
w
f
r
equenc
y
(
LF
)
ban
d (
0
.
04
-
0
.
15)
a
nd
hi
gh
f
r
equenc
y
(
H
F
)
band
(
0.
15
-
0.
4
H
z
)
r
ef
l
ec
t
t
he
t
he
s
y
m
pat
h
et
i
c
an
d
par
as
y
m
pat
h
e
t
i
c
ac
t
i
v
i
t
i
es
of
t
he A
ut
on
om
i
c
N
er
v
ous
S
y
s
t
em
(
A
N
S
)
r
es
pec
t
i
v
el
y
[
2
0]
.
In
th
i
s
s
tu
d
y
,
f
as
t
F
our
i
er
t
r
ans
f
or
m
s
(F
F
T
)
[
21]
and a
ut
o
-
r
egr
es
s
i
v
e (
A
R
)
m
odel
[
22]
w
er
e
us
ed t
o es
t
i
m
at
e t
he po
w
er
s
pec
t
r
um
f
r
o
m
H
F
and
LF
b
ands
.
R
at
i
o
of
LF
t
o
H
F
(
LF
/
H
F
)
ban
d
i
s
al
s
o
c
al
c
u
l
at
ed.
C
oef
f
i
c
i
ent
s
of
A
R
m
odel
w
er
e
es
t
i
m
at
ed w
i
t
h bur
g
m
et
hod
[
23]
an
d t
he
or
der
w
as
s
et
t
o
16
w
h
i
c
h
i
s
o
pt
i
m
al
f
or
H
R
V
ana
l
y
s
i
s
[
22]
.
B
ot
h F
F
T
and A
R
ar
e pop
ul
ar
t
oo
l
s
f
or
s
pec
t
r
al
anal
y
s
i
s
.
H
o
w
e
v
er
,
e
ac
h of
t
hem
has
t
hei
r
s
t
r
engt
hs
an
d
w
eak
nes
s
es
.
T
he
adv
ant
age
of
F
F
T
i
s
i
t
i
s
a
non
-
p
ar
am
et
r
i
c
t
ool
t
ha
t
d
oe
s
not
as
s
um
e
t
he
dat
a
i
s
u
ni
f
or
m
l
y
d
i
s
t
r
i
b
ut
ed
w
i
t
h
c
er
t
a
i
n
v
ar
i
a
nc
e
w
h
i
l
e
A
R
m
odel
as
s
um
es
t
he
dat
a
i
s
uni
f
or
m
l
y
di
s
t
r
i
b
ut
e
d
w
i
t
h f
i
x
ed
v
ar
i
anc
e
v
al
ue.
T
her
ef
or
e,
F
F
T
does
not
s
uf
f
er
f
r
o
m
poor
per
f
or
m
anc
e w
h
en t
he
pr
oper
t
y
of
dat
a
do
es
not
f
i
t
as
s
um
pt
i
on.
H
o
w
ev
er
,
F
F
T
s
u
f
f
er
f
r
o
m
s
pec
t
r
al
l
eak
age ef
f
ec
t
w
h
en c
om
par
ed t
o
A
R
m
odel
.
B
es
i
des
,
A
R
m
odel
c
an
pr
ov
i
de
bet
t
er
f
r
equenc
y
r
es
o
l
ut
i
on
i
n
po
w
er
s
pec
t
r
um
and per
f
or
m
bet
t
er
w
h
en
i
t
i
s
app
l
i
e
d t
o s
hor
t
t
i
m
e
s
er
i
es
dat
a
[
24]
.
T
her
ef
or
e,
t
hi
s
s
t
ud
y
em
pl
o
y
e
d b
ot
h
t
ec
hni
ques
f
or
s
pec
t
r
al
a
n
al
y
s
i
s
t
o t
ak
e
adv
ant
age
of
t
hei
r
s
t
r
e
ngt
h
s
.
2.
3.
3
.
B
i
sp
ect
r
u
m
F
eat
u
r
e
s
P
S
D
of
s
pec
t
r
al
an
al
y
s
i
s
does
not
pr
o
v
i
de t
h
e pha
s
e r
el
at
i
ons
bet
w
een f
r
eq
uenc
y
c
o
m
ponent
s
.
H
o
w
ev
er
,
H
i
g
her
O
r
der
S
p
ec
t
r
a (
H
O
S
)
[
25]
c
a
n b
e
us
ed
t
o
ana
l
y
z
e
t
he
no
n
-
l
i
ne
ar
s
i
gna
l
w
h
i
c
h m
a
y
i
n
v
o
l
v
e t
h
e c
r
os
s
phas
e r
el
at
i
ons
[
26
]
.
S
i
nc
e t
h
e H
R
V
s
i
gn
al
i
s
non
-
l
i
n
ear
an
d
non
-
G
aus
s
i
an
i
n
nat
ur
e,
b
i
s
pec
t
r
um
al
s
o c
an r
e
v
e
al
t
he
i
nf
or
m
at
i
on t
h
at
i
s
no
t
c
ont
a
i
n
ed
i
n
po
w
er
s
p
ec
t
r
um
]
.
B
es
i
d
es
,
t
hes
e
f
eat
ur
es
a
l
s
o c
a
n
be em
pl
o
y
ed
t
o
de
t
ec
t
q
u
adr
at
i
c
ph
as
e
c
oupl
ed
har
m
oni
c
s
ar
i
s
i
ng
f
r
o
m
nonl
i
n
ear
i
t
i
es
of
t
he
H
R
V
s
i
gna
l
.
T
he
bi
s
p
ec
t
r
um
B
(
1
,
2
)
of
a
non
-
G
aus
s
i
an
s
i
gna
l
,
(
)
,
i
s
a
t
wo
-
d
i
m
ens
i
onal
F
our
i
er
t
r
ans
f
or
m
of
t
he
t
hi
r
d
or
der
c
um
ul
ant
C
(
m
,
n
)
defi
ne
d
as
:
C
(
m
,
n
)
=
[
(
)
(
+
)
(
+
)
]
(
3)
B
(
1
,
2
)
=
[
(
1
)
∗
(
2
)
(
1
+
2
)
]
(
4)
W
h
er
e
i
s
ex
pec
t
at
i
o
n f
unc
t
i
on
,
(
)
i
s
F
our
i
er
t
r
ans
f
or
m
of
(
)
and
∗
(
)
i
s
c
om
pl
ex
c
onj
ugat
e
.
I
n
t
h
i
s
s
t
ud
y
,
B
i
s
pec
t
r
um
w
as
es
t
i
m
at
ed
bas
e
d
on
t
he
di
r
ec
t
m
et
hod
d
es
c
r
i
bed
i
n
[
25]
.
T
he 4 H
z
c
u
bi
c
s
pl
i
n
e i
nt
e
r
pol
at
ed H
R
V
s
i
gn
al
w
as
di
v
i
d
ed
i
nt
o s
ev
er
al
s
egm
ent
s
w
i
t
h e
ac
h
s
egm
ent
i
s
c
ons
i
s
t
ed of
512 dat
a po
i
nt
s
.
T
hen b
i
s
p
ec
t
r
um
w
as
t
hen c
om
put
ed f
r
o
m
F
our
i
er
t
r
ans
f
or
m
o
f
eac
h s
egm
ent
.
A
f
t
er
t
hat
,
bi
s
p
ec
t
r
um
f
eat
ur
es
w
er
e ex
t
r
ac
t
e
d f
r
o
m
di
f
f
er
ent
r
egi
ons
of
t
he t
w
o
-
d
i
m
ens
i
ona
l
b
i
ps
ec
t
r
um
.
B
i
s
pec
t
r
um
of
H
R
V
s
i
gnal
c
an be di
v
i
ded i
nt
o 3 s
ubba
nd r
eg
i
ons
i
ns
i
d
e r
egi
o
n of
i
nt
er
es
t
(
R
O
I)
[
2
7]
.
T
he
y
a
r
e LF
–
LF
(
LL)
,
LF
–
H
F
(
LH
)
,
and H
F
–
H
F
(
H
H
)
r
eg
i
on
w
h
i
c
h c
ov
er
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
V
ent
r
i
c
ul
ar
T
ac
hy
ar
r
hy
t
hm
i
a O
ns
et
P
r
e
d
ic
t
io
n
B
a
s
ed
o
n H
R
V
an
d G
en
et
i
c
…
(
K.
H
.
Bo
o
n
)
1003
di
f
f
er
ent
r
anges
of
f
r
equen
c
i
es
as
s
ho
w
n
i
n F
i
gur
e
3
.
F
or
m
ul
as
i
n
[
27,
28]
w
er
e
em
pl
o
y
ed
t
o
c
o
m
put
e bi
s
p
ec
t
r
um
f
eat
ur
es
f
r
o
m
eac
h s
ubban
d
r
eg
i
on a
nd t
he R
O
I
.
T
hes
e f
eat
ur
es
i
nc
l
ud
e
m
ean m
agni
t
u
de
(
)
,
n
or
m
al
i
z
e
d b
i
s
pec
t
r
a
l
e
nt
r
op
y
(
P
1)
,
nor
m
al
i
z
e
d b
i
s
pec
t
r
a
l
s
quar
ed
ent
r
op
y
(
P
2)
,
s
um
of
l
ogar
i
t
hm
i
c
a
m
pl
i
t
u
des
of
t
he bi
s
pec
t
r
um
(H
1
),
s
u
m
of
l
ogar
i
t
hm
i
c
am
pl
i
t
udes
of
di
ago
na
l
el
em
ent
s
i
n bi
s
pec
t
r
um
(H
2
),
f
ir
s
t
-
or
der
s
pec
t
r
al
m
om
ent
of
t
he
am
pl
i
t
udes
of
di
ag
on
al
el
e
m
ent
s
i
n
t
h
e
bi
ps
ec
t
r
um
(H
3
),
S
ec
ond
-
or
d
er
s
pec
t
r
a
l
m
o
m
ent
of
t
he
am
pl
i
t
udes
of
di
ag
on
al
e
l
e
m
ent
s
i
n t
he bi
s
pec
t
r
um
(H
4
)
,
w
ei
ght
e
d c
ent
er
of
t
he bi
ps
ec
t
r
um
,
W
CO
B
(
1
,
2
)
.
F
or
LH
r
egi
on,
H
2,
H
3
and
H
4
ar
e
ex
c
l
ud
ed
bec
a
us
e
t
h
e
d
i
ag
ona
l
e
l
em
ent
s
ar
e not
ex
i
s
t
e
d.
F
i
gur
e 3.
S
ub
-
b
and
R
e
gi
o
n
s
(
LL,
LH
and
H
H
)
an
d R
e
g
i
on
of
I
nt
er
es
t
(
R
O
I)
2.
3.
4
.
N
o
n
l
i
n
e
a
r
d
y
n
a
m
i
c
s
fe
a
tu
r
e
s
G
ener
al
l
y
,
no
n
-
l
i
near
a
nal
y
s
i
s
i
s
r
ec
ogni
z
e
d
t
o
be
abl
e
t
o
des
c
r
i
be
t
h
e
bi
o
l
og
i
c
a
l
pr
oc
es
s
i
n m
or
e ef
f
ec
t
i
v
e w
a
y
.
R
e
v
ie
w
in
[
1
9]
h
as
s
ho
w
n
t
hat
v
ar
i
o
us
non
-
lin
e
a
r
t
ec
hn
i
qu
es
hav
e
b
een ex
t
en
ded t
o
s
t
ud
y
v
ar
i
ous
c
ar
d
i
ac
ar
r
h
y
t
hm
i
as
.
I
n t
hi
s
s
t
u
d
y
,
P
o
i
nc
ar
e pl
ot
a
nd
s
a
m
pl
e e
nt
r
op
y
(
S
am
pE
n)
ar
e em
pl
o
y
e
d t
o ex
t
r
ac
t
f
ea
t
ur
es
f
r
o
m
t
he H
R
V
.
P
oi
nc
ar
e
p
l
ot
i
s
dr
a
w
n
b
y
pl
ot
t
i
n
g
eac
h
R
R
i
nt
er
v
al
a
gai
ns
t
nex
t
R
R
i
nt
er
v
a
l
.
E
a
c
h
R
R
i
nt
er
v
a
l
i
s
t
he
t
i
m
i
ng
di
f
f
er
enc
e
bet
w
e
en
s
uc
c
es
s
i
v
e
R
peak
s
o
f
H
R
V
s
i
gnal
.
A
n
el
l
i
ps
e
i
s
t
h
en
f
i
t
t
ed t
o t
h
e s
h
ape
of
P
o
i
nc
ar
e p
l
ot
.
T
he c
om
put
at
i
o
n o
f
t
he
w
i
dt
h
(
S
D
1)
an
d
l
en
gt
h (
S
D
2)
of
t
he
ec
l
i
ps
e c
an b
e s
i
m
pl
i
f
i
ed b
y
em
pl
o
y
i
ng f
or
m
ul
as
bel
o
w
.
S
t
a
ndar
d de
v
i
a
t
i
o
n de
not
e
d S
D
1 i
s
r
el
at
e
d
t
o
f
as
t
beat
-
to
-
ea
t
v
ar
i
ab
i
l
i
t
y
i
n
d
at
a
w
h
i
l
e
S
D
2
des
c
r
i
bes
t
he
l
ong
t
er
m
v
ar
i
ab
i
l
i
t
y
o
f
RR.
T
he r
at
i
o of
S
D
1/
S
D
2
i
s
al
s
o c
om
put
ed t
o
des
c
r
i
be
t
he
r
el
at
i
on
bet
w
ee
n t
w
o c
om
ponen
t
s
.
1
=
=
1
2
∗
2
(
5)
2
=
2
∗
2
−
1
2
∗
2
(
6)
S
am
pl
e
e
nt
r
op
y
(
S
am
pE
n)
i
s
a
s
t
at
i
s
t
i
c
m
eas
ur
e
t
h
at
qu
ant
i
f
i
es
t
h
e
r
eg
ul
ar
i
t
y
of
t
i
m
es
s
er
i
es
dat
a.
T
he
m
et
hod pr
opos
ed i
n
[
29]
w
as
us
ed t
o
c
o
m
put
e t
he S
am
pE
n of
t
he H
R
V
s
i
gn
al
.
P
ar
am
et
er
s
of
s
a
m
pl
e ent
r
op
y
ar
e s
et
as
f
ol
l
o
w
s
:
E
m
beddi
ng d
i
m
ens
i
on,
m
i
s
s
et
t
o 2 and
t
ol
er
a
nc
e d
i
s
t
anc
e,
r
i
s
s
et
t
o 20%
of
t
he s
t
an
dar
d d
ev
i
at
i
on
of
H
R
V
s
equ
enc
e
s
ac
c
or
di
ng t
o
r
ec
om
m
endat
i
o
n b
y
P
i
nc
us
and G
ol
d
ber
g
er
[
29]
.
S
am
pl
e ent
r
o
p
y
i
s
d
ef
i
ned as
:
(
,
)
=
l
im
→
∞
(
−
1
−
∑
−
=
1
1
−
∑
−
=
1
)
(
7)
W
h
er
e,
(
.
)
i
s
H
ea
v
i
s
i
de f
unc
t
i
on a
nd,
=
1
−
∑
(
−
|
|
+
1
(
)
−
+
1
(
)
|
|
)
−
=
1
,
≠
(
8)
=
1
−
∑
(
−
|
|
(
)
−
(
)
|
|
)
−
=
1
,
≠
(
9)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
99
–
1
0
08
1004
2.
4
.
C
l
a
s
s
i
fi
c
a
ti
o
n
:
S
u
p
p
o
r
t v
e
c
to
r
m
a
c
h
i
n
e
(
S
VM
)
In
th
i
s
s
tu
d
y
,
SV
M
i
s
us
ed
as
s
uper
v
i
s
e
d
c
la
s
s
if
ie
r
to
c
l
as
s
i
f
y
t
h
e H
R
V
s
eq
uenc
es
to
ei
t
h
er
“
nor
m
al
epi
s
o
de”
(
c
ont
r
ol
dat
a
)
or
“
abn
or
m
al
epi
s
o
de”
(
H
R
V
s
e
qu
enc
es
pr
i
or
t
o
V
T
A
ev
e
nt
)
.
I
nput
of
t
h
e S
V
M
i
s
t
he H
R
V
f
ea
t
ur
es
t
h
at
ar
e
ex
t
r
ac
t
ed
d
ur
i
ng
t
h
e f
eat
ur
e ex
t
r
ac
t
i
o
n
s
t
age.
S
V
M
i
s
c
hos
en b
ec
aus
e s
i
m
i
l
ar
H
R
V
b
as
ed r
el
at
ed
H
R
V
an
al
y
s
i
s
bas
e
d
w
or
k
s
[
30]
as
r
epor
t
ed
go
od c
l
as
s
i
f
i
c
at
i
o
n
per
f
or
m
anc
e w
i
t
h
t
h
i
s
c
l
as
s
i
f
i
er
.
SV
M
i
s
s
u
per
v
i
s
ed
c
l
as
s
i
f
i
er
bas
ed
o
n s
t
at
i
s
t
i
c
a
l
l
ear
ni
n
g t
h
eor
y
[
3
1]
.
S
V
M m
aps
t
he
t
r
ai
n
i
ng
s
am
pl
es
f
r
o
m
t
he
i
npu
t
s
pac
e
t
o
h
i
g
her
-
di
m
ens
i
onal
f
eat
ur
es
s
pac
e
v
i
a
a
k
er
nel
f
unc
t
i
on.
I
n t
h
i
s
s
t
ud
y
,
t
he r
adi
al
bas
i
s
f
unc
t
i
o
n (
R
B
F
)
i
s
us
ed as
t
h
e k
er
nel
f
unc
t
i
o
n.
P
ar
am
e
t
e
r
s
of
k
er
nel
-
k
er
nel
w
i
dt
h
γ
an
d
r
egul
ar
i
z
at
i
o
n c
ons
t
ant
C
–
ar
e
s
et
ac
c
or
di
n
g
t
o f
ea
t
ur
e
s
el
ec
t
i
on
m
et
hod i
n s
ec
t
i
on
2.
5.
2.
5
. G
e
n
e
t
ic
A
lg
o
r
it
h
m
B
a
sed
F
eat
u
r
e
S
el
e
ct
i
o
n
T
he pur
pos
e of
f
eat
ur
e s
el
ec
t
i
on i
s
t
o s
el
ec
t
opt
i
m
al
s
ubs
et
of
f
eat
ur
es
f
r
o
m
or
i
g
in
a
l
f
eat
ur
e
s
et
w
i
t
ho
ut
t
r
ans
f
or
m
s
or
i
gi
nal
f
eat
ur
e.
U
s
i
ng
al
l
t
h
e
ex
t
r
ac
t
ed
f
eat
ur
es
d
oes
not
al
w
a
y
s
gi
v
e
t
he
b
es
t
c
l
as
s
i
f
i
c
at
i
on
per
f
or
m
anc
e
[
32]
.
F
eat
ur
e
s
el
ec
t
i
on
pr
oc
es
s
c
an
of
f
er
t
w
o
b
enef
i
t
s
:
enha
nc
e c
l
as
s
i
f
i
c
at
i
on
per
f
or
m
anc
e and r
ed
uc
e num
ber
of
f
eat
ur
es
r
eq
ui
r
e
d f
or
c
l
as
s
i
f
i
c
at
i
on
m
odel
.
F
ur
t
her
m
or
e,
i
t
al
s
o
c
an
hel
p
us
t
o
und
er
s
t
and
w
h
i
c
h
f
eat
ur
es
ar
e
i
m
por
t
ant
f
or
c
la
s
s
if
ic
a
t
io
n
.
I
n our
w
or
k
,
f
eat
ur
e s
el
ec
t
i
on pr
oc
es
s
bas
ed o
n G
en
et
i
c
A
l
gor
i
t
hm
(
G
A
)
pr
opos
ed b
y
H
uan
g
an
d
W
ang
[
15]
i
s
a
dopt
ed
t
o
s
i
m
ul
t
ane
ous
l
y
o
pt
i
m
i
z
e
t
he
H
R
V
f
eat
ur
e
s
u
bs
et
an
d
S
V
M
par
am
et
er
s
(
C
and
γ
)
.
I
n
i
t
i
al
l
y
,
G
A
pr
o
duc
es
a
n
i
ni
t
i
a
l
pop
ul
a
t
i
o
n
w
i
t
h
s
i
z
e
of
N
c
hr
om
os
o
m
e
s
.
E
ac
h c
hr
om
os
o
m
e i
s
r
epr
es
ent
ed b
y
f
i
x
ed l
eng
t
h b
i
na
r
y
s
t
r
i
n
g.
B
i
nar
y
s
t
r
i
ng c
an be di
v
i
ded i
nt
o
3 s
egm
ent
s
.
F
i
r
s
t
s
eg
m
ent
i
s
53 bi
t
s
bi
nar
y
s
t
r
i
ng t
h
at
r
epr
es
ent
s
a f
eat
ur
e s
ubs
et
,
s
uc
h t
hat
“
1”
r
epr
es
ent
s
s
e
l
ec
t
i
on
w
hi
l
e
‘
‘
0
’
’
r
epr
es
e
nt
s
t
h
e d
el
e
t
i
o
n of
t
he
s
pec
i
f
i
c
f
eat
ur
e f
r
om
t
he f
eat
ur
e
s
et
.
S
ec
o
nd
and t
hi
r
d
s
eg
m
ent
ar
e 20 b
i
t
s
bi
nar
y
s
t
r
i
ng t
hat
r
epr
es
ent
s
t
he
en
c
oded
v
a
l
ue
of
par
am
et
er
C
and
par
am
et
e
r
γ
f
or
S
V
M
r
es
pec
t
i
v
el
y
.
B
i
nar
y
s
t
r
i
ng
of
s
ec
ond
and
t
hi
r
d
s
egm
ent
i
s
dec
od
ed
bac
k
t
o r
eal
v
a
l
ue
w
i
t
h
equ
at
i
on 9
:
=
+
−
2
−
1
×
(
10)
W
h
er
e
i
s
r
eal
v
a
l
ue
of
t
he
bi
n
ar
y
s
t
r
i
n
g
,
is
d
e
c
im
a
l
o
f
b
it
s
t
r
in
g
,
i
s
m
a
x
i
m
u
m
v
a
l
ue
of
par
am
et
er
,
i
s
m
i
ni
m
u
m
v
a
l
ue
of
par
am
et
er
,
l
i
s
l
en
gt
h
of
bi
t
s
t
r
i
ng.
I
n
t
h
i
s
pa
per
,
bot
h
and
ar
e s
et
t
o 0.
1 a
nd
100
0 r
es
pec
t
i
v
e
l
y
f
or
bot
h
S
V
M
par
am
et
er
s
.
F
i
t
nes
s
f
unc
t
i
o
n
i
n
e
quat
i
o
n
(
11)
i
s
us
e
d
t
o
ev
al
uat
e
fi
t
nes
s
of
a
c
hr
om
os
o
m
e.
=
×
+
∑
×
=
1
−
1
(
11)
W
h
er
e
i
s
w
ei
ght
f
or
S
V
M pr
e
di
c
t
i
on
ac
c
ur
ac
y
,
i
s
w
ei
g
ht
f
or
s
el
ec
t
ed f
eat
ur
es
,
i
s
pr
edi
c
t
i
on
ac
c
ur
ac
y
of
S
V
M,
i
s
co
st
o
f
i
th
f
eat
ur
e,
r
epr
es
ent
s
t
he
w
het
her
i
t
h
f
eat
ur
e i
s
s
e
l
ec
t
e
d or
not
.
I
n t
h
i
s
pa
per
,
an
d
ar
e s
et
t
o 0.
8
an
d 0
.
2 r
es
p
ec
t
i
v
e
l
y
ac
c
or
di
ng
t
o ex
p
er
i
m
ent
s
et
t
i
ng
i
n H
u
ang
an
d
W
ang
[
15]
.
B
e
s
id
e
s
,
a
l
l
ar
e s
e
t
t
o “
1”
s
i
nc
e
t
h
e i
nt
er
es
t
of
f
eat
ur
e
s
el
ec
t
i
on
i
n
o
ur
w
or
k
i
s
t
o
m
i
ni
m
i
z
e
f
eat
ur
e
c
ount
dur
i
ng
t
he
o
pt
i
m
i
z
a
t
i
o
n
(
not
t
o
r
educ
e
f
eat
ur
e c
om
put
at
i
o
n c
os
t
or
dol
l
ar
c
os
t
)
.
R
oul
et
t
e
W
heel
S
e
l
ec
t
i
on
m
et
hod i
s
us
ed
as
a
s
el
ec
t
i
on s
t
r
at
eg
y
.
D
oub
l
e
po
i
n
t
c
r
os
s
ov
er
oper
at
or
and
bi
t
f
l
i
p m
ut
at
i
on ar
e
em
pl
o
y
ed as
ge
net
i
c
oper
at
or
s
t
o ex
pl
or
e t
h
e
s
ear
c
h
s
pac
e.
O
t
her
par
a
m
et
er
s
of
G
A
ar
e
s
et
as
f
ol
l
o
w
s
:
P
r
oba
bi
l
i
t
y
of
c
r
os
s
ov
er
(
P
c
)
=
0.
7
,
P
r
oba
bi
l
i
t
y
of
m
ut
at
i
on
(
P
m
)
=
0.
01
,
S
t
op
pi
ng
G
ener
at
i
o
n
(
G
N
)
=
500
0
a
nd
po
pul
at
i
on
s
i
z
e
(
N
)
=6
0
.
2.
6
.
P
e
r
fo
r
m
a
n
c
e
E
v
a
l
u
a
ti
o
n
I
n
t
h
i
s
s
t
ud
y
,
S
V
M
c
l
as
s
i
f
i
e
r
i
s
us
ed
t
o
ev
al
uat
e
t
h
e
pr
edi
c
t
i
on
per
f
or
m
anc
e
of
s
el
ec
t
ed
H
R
V
f
eat
ur
es
i
n
t
h
e
f
eat
ur
e
s
el
ec
t
i
o
n
pr
oc
es
s
.
P
er
f
or
m
anc
e
m
et
r
i
c
s
s
uc
h
as
s
ens
i
t
i
v
i
t
y
(
S
E
N
)
,
s
pec
i
fi
c
i
t
y
(
S
P
E
)
,
an
d
ac
c
ur
ac
y
(
A
C
C
)
,
w
h
i
c
h
h
av
e
be
en
us
ed
i
n
[
3
3]
,
ar
e
em
pl
o
y
ed t
o
m
eas
ur
e
t
he pr
ed
i
c
t
i
on
per
f
or
m
anc
e
of
al
gor
i
t
hm
.
P
os
i
t
i
v
e
pr
e
d
i
c
t
i
o
n m
eans
t
he
al
gor
i
t
hm
c
l
as
s
i
f
i
es
t
he
R
R
i
nt
er
v
a
l
s
r
ec
or
di
n
g as
pr
i
or
t
o P
A
F
ev
ent
c
or
r
e
c
t
l
y
w
h
i
l
e neg
at
i
v
e
pr
edi
c
t
i
on m
eans
t
he
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
V
ent
r
i
c
ul
ar
T
ac
hy
ar
r
hy
t
hm
i
a O
ns
et
P
r
e
d
ic
t
io
n
B
a
s
ed
o
n H
R
V
an
d G
en
et
i
c
…
(
K.
H
.
Bo
o
n
)
1005
al
g
or
i
t
hm
c
l
as
s
i
f
i
es
t
he
R
R
i
nt
er
v
al
s
r
ec
or
d
i
ng
f
r
om
c
ont
r
ol
d
at
a
c
or
r
ec
t
l
y
.
T
he
s
ens
i
t
i
v
i
t
y
(
S
E
N
)
i
s
defi
n
ed
as
t
he
r
at
i
o
of
t
he
n
um
ber
c
or
r
ec
t
pos
i
t
i
v
e
pr
ed
i
c
t
i
o
n
t
o
t
h
e
t
ot
a
l
n
um
ber
of
pos
i
t
i
v
e
pr
edi
c
t
i
on
.
S
p
ec
i
fi
c
i
t
y
(
S
P
E
)
i
s
t
he
r
at
i
o
of
t
he
num
ber
of
c
or
r
ec
t
negat
i
v
e
pr
edi
c
t
i
on
t
o
t
ot
a
l
num
ber
of
negat
i
v
es
pr
e
d
i
c
t
i
on.
A
c
c
ur
ac
y
(
A
C
C
)
i
s
t
he r
at
i
o
of
t
ot
a
l
n
um
ber
of
c
or
r
ec
t
pr
edi
c
t
i
on
.
I
n
c
ur
r
ent
w
or
k
s
,
50%
of
s
a
m
pl
e
dat
a
ar
e
r
an
do
m
l
y
s
el
ec
t
ed
as
t
r
a
i
ni
ng
s
et
and
r
em
ai
ni
ng
dat
a ar
e
us
e
d
as
t
es
t
i
ng s
e
t
.
F
ur
t
her
m
or
e,
s
am
pl
e d
at
a
f
or
bot
h t
r
ai
n
i
ng
s
et
an
d
t
es
t
i
n
g ar
e
s
ubj
ec
t
e
d
i
nd
e
pend
ent
af
t
er
t
h
e
y
ar
e
r
an
dom
l
y
par
t
i
t
i
on
ed.
T
her
ef
or
e,
R
R
i
nt
er
v
al
r
ec
or
di
ngs
f
r
om
bot
h t
r
ai
ni
n
g s
et
a
nd t
es
t
i
n
g s
et
ar
e de
f
i
ni
t
e
l
y
c
om
e f
r
o
m
di
f
f
er
ent
pat
i
ent
s
.
3.
R
e
s
u
l
t a
n
d
D
i
scu
ss
i
o
n
s
T
abl
e 1 s
ho
w
s
t
he be
nc
h
m
ar
k
i
ng r
es
ul
t
s
of
our
pr
o
pos
ed m
et
hod ag
ai
ns
t
pr
e
v
i
ous
w
or
k
s
.
T
he benc
h
m
ar
k
i
ng as
pec
t
s
i
nc
l
u
de t
h
e r
equ
i
r
e
d H
R
V
s
i
g
na
l
l
en
gt
h f
or
pr
edi
c
t
i
on,
t
y
pe of
H
R
V
f
eat
ur
e ex
t
r
ac
t
i
on m
et
hod,
per
f
or
m
anc
e ev
al
uat
i
o
n m
et
hod an
d pr
e
di
c
t
i
on
pe
r
f
or
m
anc
e.
I
n T
abl
e
1,
t
he r
es
ul
t
s
h
a
v
e s
h
o
w
n
t
hat
t
h
e pr
e
di
c
t
i
on
per
f
or
m
anc
e of
our
m
et
ho
d
out
p
er
f
or
m
s
al
l
pr
e
v
i
o
us
w
or
k
s
.
W
i
t
h 5
m
i
nut
es
of
H
R
V
s
i
gn
al
l
eng
t
h pr
i
or
t
o V
T
A
ons
et
,
our
m
et
hod ac
hi
e
v
es
7
9.
4
1%
o
f
ac
c
ur
ac
y
,
w
h
i
c
h
i
s
hi
gher
t
han t
he
ac
c
ur
ac
y
r
ep
or
t
ed
b
y
J
oo
,
et
al
.
,
[
11]
(
bes
t
pr
ev
i
o
us
w
or
k
)
and W
ol
l
m
an
,
e
t
a
l.
,
[
1
4]
r
es
pec
t
i
v
e
l
y
.
A
l
t
ho
ugh
t
he ac
c
ur
ac
y
le
v
e
ls
w
er
e
not
r
e
por
t
e
d
b
y
T
hon
g
&
R
a
i
t
t
[
1
2]
a
nd
R
oz
en
,
et
al
.
,
[
13]
,
t
he
i
r
pr
ed
i
c
t
i
o
n
s
ens
i
t
i
v
i
t
y
a
nd
s
pec
i
f
i
c
i
t
y
w
er
e
not
ba
l
anc
e
d
w
i
t
h
l
o
w
s
e
ns
i
t
i
v
i
t
y
r
at
e
(
5
3 an
d 5
0%
)
.
I
t
s
how
s
t
ha
t
t
h
e
i
r
p
r
e
di
c
t
i
on
m
et
hods
ac
hi
ev
e
d
poor
pe
r
f
or
m
anc
e
i
n
pr
edi
c
t
i
n
g
t
he
V
T
A
ons
et
s
uc
c
es
s
f
ul
l
y
.
I
n
c
ont
r
as
t
,
our
m
et
hod
c
an
ac
h
i
e
v
e
ac
c
ep
t
abl
e
a
nd
ba
l
anc
e
d
pr
e
di
c
t
i
on
s
ens
i
t
i
v
i
t
y
an
d
s
pec
i
f
i
c
i
t
y
w
i
t
h
77.
94%
and 80
.
88%
r
es
p
ec
t
i
v
el
y
.
S
i
m
i
l
ar
bal
a
nc
ed pr
e
di
c
t
i
on s
ens
i
t
i
v
i
t
y
and s
pec
i
f
i
c
i
t
y
ar
e
onl
y
r
e
por
t
e
d
b
y
Jo
o
,
et
al
.
,
[
1
1]
.
T
abl
e 1
.
Be
n
c
hm
ar
k
i
ng aga
i
ns
t
pr
e
v
i
ous
w
or
k
s
P
r
evi
o
u
s
W
o
rk
HRV
S
i
g
n
a
l
Le
ngt
h
(M
i
n
u
te
s
)
F
eat
u
r
e E
xt
r
act
i
o
n
M
e
t
ho
d
P
er
f
o
r
m
a
n
ce E
va
l
u
at
i
o
n
M
e
t
ho
d
SEN
(
%
)
SP
E
(
%
)
A
C
C
(
%
)
T
hong &
R
ai
t
t
,
2007.
[
12]
1.
8
H
our
s
D
ec
i
s
i
on
r
ul
e ba
s
ed
on H
R
pat
t
er
n.
U
s
ed al
l
208 dat
a
as
bot
h
t
r
ai
ni
ng
and t
es
t
i
ng s
et
.
53.
0
91.
0
-
R
o
z
en
et
al
.
,
2013
[
13]
10
-
60
D
ec
i
s
i
on
r
ul
e ba
s
ed
on M
ul
t
i
pol
e
anal
y
s
i
s
.
U
s
ed al
l
124 dat
a
as
bot
h
t
r
ai
ni
ng
and t
es
t
i
ng s
et
.
(
64
p
re
-
V
T
/
V
F
,
60
c
on
t
r
ol
dat
a)
50.
0
91.
6
-
W
ol
l
m
a
n e
t
al
.
,
2015
[
14]
20
-
40
T
i
m
e do
m
ai
n
f
eat
ur
es
.
(
CA
RT
)
U
s
ed al
l
155 dat
a
as
bot
h
t
r
ai
ni
ng
and t
es
t
i
ng s
et
.
(
68
p
re
-
V
T
/
V
F
,
72
c
on
t
r
ol
dat
a)
94.
4
50.
6
70.
9
Jo
o
et
a
l
.
,
2012.
[
11]
5
T
i
m
e,
W
el
c
h
ba
s
ed
F
F
T
,
P
oi
n
c
ar
e.
A
N
N
P
ar
t
i
t
i
oned t
he
dat
abas
e
i
nt
o
175 t
r
ai
ni
ng dat
a
and
86
t
es
t
i
ng dat
a
.
77.
3
(
34/
44)
73.
8
(
31/
42)
75.
6
(
65/
86)
Ou
r
pr
opos
e
d
M
e
t
ho
d
5
T
i
m
e
, F
F
T
, A
R
,
P
oi
nc
ar
e,
H
i
gher
O
r
der
S
pec
t
r
al
.
SVM
.
P
ar
t
i
t
i
oned t
he
dat
abas
e
i
nt
o
134 t
r
ai
ni
ng dat
a
and 135
t
es
t
i
ng dat
a
.
77.
94
(
53/
68)
80.
88
(
55/
68)
79.
41
(
108/
136)
I
n t
er
m
o
f
per
f
or
m
anc
e ev
al
uat
i
on m
et
hod,
di
r
ec
t
c
om
p
ar
i
s
on c
an be m
ade bet
w
e
en our
m
et
hod and t
he b
es
t
pr
e
di
c
t
i
on m
et
hod
pr
opos
e
d b
y
J
oo
,
et
al
.
,
[
11]
bec
a
us
e T
abl
e 1
s
ho
w
s
t
hat
b
ot
h
of
our
w
or
k
s
us
e
i
nde
pen
de
nt
t
r
ai
ni
n
g
an
d
t
es
t
i
ng
dat
a
s
e
t
t
o
t
r
a
i
n an
d
ev
al
u
at
e t
he
pr
opos
e
d m
et
hod.
O
t
her
pr
ev
i
ous
w
or
k
s
[
12
-
14]
d
i
d
n
ot
p
er
f
or
m
s
uc
h par
t
i
t
i
on
i
ng
.
I
ns
t
e
ad,
t
he
i
r
t
r
ai
n
i
ng a
nd t
es
t
i
ng s
et
s
ha
r
ed t
he s
am
e s
a
m
pl
e dat
a.
I
n our
w
or
k
,
t
he appr
oac
h t
o ev
a
l
u
at
e t
h
e
pr
edi
c
t
i
on
per
f
or
m
anc
e
of
t
he
pr
op
os
ed
m
et
hod
i
s
s
t
r
i
c
t
er
t
han
i
n
[
11
]
.
F
i
r
s
t
l
y
,
h
i
g
her
num
ber
of
t
es
t
i
n
g
s
am
pl
e
da
t
a
i
s
us
e
d
t
o
m
eas
ur
e
t
h
e
pr
e
di
c
t
i
o
n
ac
c
ur
ac
y
.
W
e
hav
e
us
ed
50%
of
s
a
m
pl
e
dat
a as
t
r
ai
n
i
n
g s
et
w
h
i
l
e r
em
ai
ni
ng 50
%
of
dat
a ar
e us
ed as
t
es
t
i
ng s
et
.
I
n c
ont
r
as
t
,
J
oo
,
et
al
.
,
[
11]
us
e
d
6
7%
of
dat
a
f
or
t
r
ai
ni
ng
and
on
l
y
32.
9%
of
dat
a
f
or
t
es
t
i
ng.
T
her
ef
or
e,
t
he
n
um
ber
of
s
a
m
pl
e
dat
a
t
o
t
es
t
t
he
t
r
ai
ned
s
uper
v
i
s
e
d
S
V
M
c
l
as
s
i
f
i
er
i
n
our
w
or
k
i
s
i
nc
r
eas
ed
b
y
62%
(
26
s
a
m
pl
es
)
.
S
ec
on
dl
y
,
s
am
pl
e dat
a f
or
bot
h t
r
ai
ni
n
g
and t
es
t
i
ng s
et
i
n our
w
o
r
k
ar
e s
ubj
ec
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
99
–
1
0
08
1006
i
nd
epe
nde
nt
af
t
er
d
at
a
ar
e
r
andom
l
y
p
ar
t
i
t
i
on
ed.
C
ons
eque
nt
l
y
,
t
he
R
R
i
nt
er
v
a
l
r
e
c
or
di
ngs
f
r
om
bot
h
t
r
ai
ni
ng
and
t
es
t
i
ng
s
et
ar
e
def
i
ni
t
el
y
o
bt
a
i
n
ed
f
r
om
di
f
f
er
ent
pa
t
i
e
nt
s
.
T
hi
s
appr
o
ac
h
w
a
s
not
r
epor
t
e
d
i
n
[
1
1]
and
t
h
e
y
onl
y
s
t
at
e
d
t
he
dat
a
ar
e
r
andom
l
y
di
v
i
de
d
i
nt
o
t
r
ai
n
i
n
g
and
t
es
t
i
ng
s
et
.
W
i
t
h s
t
r
i
c
t
er
per
f
or
m
anc
e e
v
al
uat
i
o
n appr
oac
h
,
our
pr
op
os
ed m
et
hod a
c
hi
e
v
es
hi
gher
pr
edi
c
t
i
on
ac
c
ur
ac
y
w
i
t
h
79
.
41%
t
han m
et
hod
i
n
[
11
]
(
75.
6%
ac
c
ur
ac
y
)
.
H
i
gh
er
pr
ed
i
c
t
i
o
n
per
f
or
m
anc
e
of
our
m
et
hod
c
an
be
at
t
r
i
b
ut
ed
t
o
em
pl
o
y
m
ent
o
f
m
o
r
e
t
y
p
es
of
c
o
m
pr
ehens
i
v
e
H
R
V
f
eat
ur
es
.
O
ur
pr
op
os
ed
m
et
hod
us
es
t
h
e
H
R
V
f
ea
t
ur
es
ex
t
r
ac
t
ed
f
r
o
m
s
a
m
pl
e ent
r
op
y
,
h
i
g
h
er
or
der
s
pec
t
r
al
ana
l
y
s
i
s
(
H
O
S
)
and
T
r
i
angu
l
ar
I
nt
er
pol
at
i
on of
N
N
i
nt
er
v
a
l
h
i
s
t
ogr
am
(
T
I
N
N
)
t
hat
ar
e
no
t
us
e
d
i
n
[
1
1]
.
F
ur
t
her
m
or
e,
opt
i
m
al
f
eat
ur
e
s
ubs
et
i
s
al
s
o
opt
i
m
i
z
e
d b
y
gen
et
i
c
a
l
go
r
i
t
hm
(
G
A
)
t
o r
educ
e t
he f
eat
ur
e c
ount
of
opt
i
m
al
f
eat
ur
e s
ubs
e
t
.
O
pt
i
m
al
f
eat
ur
e s
ubs
et
s
el
ec
t
ed b
y
G
A
c
ont
ai
ns
f
ol
l
o
w
i
n
g f
eat
ur
es
:
m
eanR
R
,
S
D
N
N
,
N
N
50,
H
R
V
T
r
i
,
s
am
pl
e e
nt
r
op
y
,
S
D
2,
r
at
i
o
of
S
D
1
t
o
S
D
2
,
l
o
w
f
r
eque
nc
y
ba
nd
en
e
r
g
y
of
F
F
T
,
2
bi
s
pec
t
r
um
f
eat
ur
es
f
r
o
m
LL
r
egi
on
(
P
2
H
1)
,
3
bi
s
pec
t
r
um
f
eat
ur
es
f
r
o
m
LH
r
egi
o
n
(
P
1,
P
2
a
nd
W
CO
B
(
2
)
.
)
,
2
bi
s
pec
t
r
um
f
eat
ur
es
of
H
H
r
egi
on
(
and
H
2)
,
5
bi
s
pec
t
r
um
f
eat
ur
es
f
r
o
m
R
O
I
r
egi
on (
P
1,
W
C
O
B (
1
)
.
,
H
2,
H
3
an
d H
4)
.
A
not
her
ad
v
a
nt
a
ge
of
our
m
et
hod
i
s
us
i
ng
s
hor
t
er
H
R
V
s
i
g
nal
l
en
gt
h
dur
i
n
g
t
h
e
f
eat
ur
e
ex
t
r
ac
t
i
on.
T
abl
e
1
s
h
o
w
s
t
hat
o
ur
m
et
hod
on
l
y
us
es
5
m
i
nut
es
of
H
R
V
s
i
g
nal
l
e
ngt
h
t
hat
en
d
i
m
m
edi
at
el
y
pr
i
or
t
o
V
T
A
o
ns
et
t
o
pr
ed
i
c
t
t
h
e
V
T
A
o
n
s
et
.
I
n
c
o
nt
r
as
t
,
a
l
m
os
t
a
l
l
pr
ev
i
ous
w
or
k
s
ex
c
ept
[
11]
us
e
d
m
or
e
t
h
an
10
m
i
nut
es
of
H
R
V
s
i
gna
l
f
or
pr
edi
c
t
i
on
.
R
o
ze
n
,
et
al
.
,
[
1
3]
per
f
or
m
ed m
ul
t
i
pol
e a
na
l
y
s
i
s
on
10
t
o
60 m
i
nut
es
of
H
R
V
s
i
gna
l
.
W
o
llm
a
n
,
e
t
a
l.
,
[
14]
ex
t
r
ac
t
ed
t
hei
r
H
R
V
t
i
m
e
do
m
ai
n
f
eat
ur
es
f
r
o
m
20
t
o
40
m
i
nut
es
H
R
V
s
i
gn
al
.
F
i
na
l
l
y
,
T
ho
ng
&
R
ai
t
t
[
12]
i
de
nt
i
f
i
e
d t
he H
R
pat
t
er
n i
n
1.
8 ho
ur
s
of
s
i
gnal
f
or
pr
edi
c
t
i
on.
L
ong
dur
at
i
on of
H
R
V
s
i
gn
al
l
eng
t
h
f
or
V
T
A
ons
e
t
pr
edi
c
t
i
on
c
aus
es
s
ev
er
al
d
i
s
ad
v
a
nt
ag
es
w
hen t
he pr
ed
i
c
t
i
on m
et
hod i
s
i
m
pl
em
ent
ed i
n
bat
t
er
y
p
o
w
er
ed
I
C
D
de
v
i
c
e.
F
i
r
s
t
l
y
,
l
on
ger
d
ur
at
i
on
of
i
np
u
t
(
H
R
V
s
i
gn
al
)
i
nt
r
od
uc
es
l
o
ng
er
pr
oc
es
s
i
ng t
i
m
e i
n t
he f
eat
ur
e ex
t
r
a
c
t
i
on s
t
a
ge
w
hi
c
h m
a
y
pr
o
v
e pr
oh
i
b
i
t
i
v
e
in
r
eal
-
t
i
m
e pr
ed
i
c
t
i
o
n a
nd t
er
m
i
nat
i
on
of
V
T
A
ons
et
.
F
ur
t
her
m
or
e,
i
n r
ec
e
nt
y
e
ar
s
,
m
an
y
r
es
ear
c
hes
[
34
-
3
6]
h
av
e b
een p
er
f
or
m
ed t
o ad
dr
es
s
t
he p
o
w
er
c
ons
um
pt
i
o
n i
s
s
ue i
n
I
C
D
or
s
i
m
i
l
ar
dev
i
c
es
t
ha
t
us
e H
R
V
an
al
y
s
i
s
f
or
r
eal
t
i
m
e di
s
eas
e d
i
ag
nos
i
s
.
I
n t
h
e c
as
e
of
V
T
A
ons
et
pr
edi
c
t
i
on m
et
hods
,
t
he m
ai
n c
onc
er
n
i
s
t
hat
l
o
ng d
ur
a
t
i
on
of
s
i
gna
l
and c
om
put
e
-
i
nt
ens
i
v
e H
R
V
ana
l
y
s
i
s
a
l
g
or
i
t
hm
s
m
a
y
b
u
r
den
t
h
e
I
C
D
bat
t
er
y
l
i
f
e,
a
nd
c
ons
eq
ue
nt
l
y
s
hor
t
eni
n
g
i
t
s
op
er
at
i
on
t
i
m
e.
T
hi
s
m
a
y
c
aus
e
hi
g
h
er
f
r
equenc
y
of
bod
y
s
ur
g
er
y
pr
oc
es
s
es
t
o
r
epl
ac
e t
he
I
C
D
bat
t
er
y
,
w
hi
c
h c
an af
f
ec
t
t
he heal
t
h of
t
he pat
i
e
nt
[
34
]
.
(
G
ener
a
l
l
y
,
t
h
e I
C
D
de
v
i
c
e i
s
ex
pec
t
ed t
o op
er
at
e
f
or
m
or
e t
han
5
y
e
ar
s
af
t
er
i
t
i
s
i
m
pl
ant
e
d
i
n t
he
hu
m
an bod
y
)
.
T
her
ef
or
e,
s
hor
t
er
H
R
V
s
i
g
na
l
l
en
gt
h
i
n
f
eat
ur
e
ex
t
r
ac
t
i
on
s
t
age
c
an
r
educ
e
bo
t
h
t
he
t
i
m
e
l
ag
b
et
w
e
en
i
n
put
s
i
g
nal
and
ou
t
put
pr
edi
c
t
i
on
,
an
d t
h
e b
ur
den
t
o t
he
bat
t
er
y
of
el
ec
t
r
o
ni
c
d
ev
i
c
e.
4
.
C
o
n
c
l
u
s
i
o
n
I
n t
hi
s
p
aper
,
a
v
e
nt
r
i
c
u
l
ar
t
ac
h
y
ar
r
h
y
t
hm
i
a (
V
T
A
)
on
s
et
pr
edi
c
t
i
on m
et
hod b
as
ed on
H
R
V
a
na
l
y
s
i
s
and
G
A
i
s
p
r
opos
ed.
W
i
t
h addi
t
i
on
al
t
y
pe of
H
R
V
f
eat
ur
es
a
nd o
pt
i
m
i
z
at
i
on
b
y
G
A
,
w
e
h
av
e
s
ho
w
n
t
hat
pr
ed
i
c
t
i
o
n
ac
c
ur
ac
y
of
p
r
opos
ed
V
T
A
o
ns
et
pr
ed
i
c
t
i
on
m
et
hod
out
p
er
f
or
m
s
al
l
pr
e
v
i
ous
w
or
k
s
ev
en
w
i
t
h
s
t
r
i
c
t
er
per
f
or
m
anc
e ev
al
u
at
i
on
a
ppr
oac
h.
T
he
pr
edi
c
t
i
on s
ens
i
t
i
v
i
t
y
a
nd
s
pec
i
f
i
c
i
t
y
of
our
m
et
hod ar
e al
s
o ac
c
ept
ab
l
e an
d
bal
anc
ed
w
hen
c
o
m
par
ed
t
o
pr
ev
i
ous
w
or
k
s
[
12
-
14]
.
B
es
i
des
,
t
he
s
el
e
c
t
ed
opt
i
m
al
f
eat
ur
e
s
et
i
s
al
s
o
r
epor
t
e
d.
F
ur
t
her
m
or
e,
our
m
et
hod
onl
y
r
e
qu
i
r
es
5
m
i
nut
es
H
R
V
s
i
gn
al
i
n l
eng
t
h,
w
h
i
c
h
i
s
s
hor
t
er
t
han
m
o
s
t
of
t
he pr
e
v
i
o
us
w
or
k
s
,
f
or
t
he ac
c
ur
at
e
pr
ed
i
c
t
i
o
n.
A
s
f
or
l
i
m
i
t
at
i
o
n,
pr
ed
i
c
t
i
o
n
r
es
ul
t
s
of
pr
opos
ed pr
edi
c
t
i
on m
et
hod ar
e l
i
m
i
t
ed b
y
s
m
al
l
s
a
m
pl
e
s
i
z
e
(
27
0
i
n
t
o
t
al
)
of
r
eal
dat
a
f
r
o
m
pa
t
i
ent
s
al
t
ho
ugh
w
e
hav
e
us
ed
h
i
gh
es
t
num
ber
o
f
s
a
m
pl
e d
at
a
w
hen
c
om
par
ed t
o
pr
e
v
i
ous
w
or
k
s
.
T
he
r
ef
or
e,
r
es
ul
t
s
m
a
y
s
uf
f
er
f
r
om
a l
ac
k
of
s
t
a
t
is
t
ic
a
l s
a
m
p
lin
g
f
or
V
T
A
pat
i
ent
.
S
om
e f
ut
ur
e w
or
k
s
c
an be don
e t
o
ex
t
en
d c
ur
r
ent
w
or
k
.
F
i
r
s
t
l
y
,
t
h
e op
t
i
m
al
v
al
ue
of
par
am
et
er
f
or
s
o
m
e
f
eat
ur
e ex
t
r
ac
t
i
o
n t
ec
h
ni
ques
c
a
n be
i
n
v
es
t
i
gat
ed t
o i
m
pr
ov
e pr
edi
c
t
i
on
per
f
or
m
anc
e of
H
R
V
f
eat
u
r
es
.
T
hes
e par
am
et
er
s
i
nc
l
ud
e or
d
er
of
A
R
r
e
gr
es
s
i
v
e,
em
bedd
i
ng
di
m
ens
i
on an
d t
o
l
er
anc
e o
f
s
a
m
pl
e ent
r
op
y
,
w
i
n
do
w
f
unc
t
i
on of
F
F
T
and et
c
.
Mor
e c
om
pl
ex
s
uper
v
i
s
ed c
l
as
s
i
f
i
er
s
uc
h
a
s
bac
k
pr
opagat
i
o
n n
eur
a
l
net
w
or
k
(
B
P
N
N
)
al
s
o
c
an
b
e em
pl
o
y
e
d t
o
r
epl
ac
e
t
he
S
V
M f
or
de
v
e
l
o
pm
ent
of
bet
t
er
s
uper
v
i
s
e
d
pr
edi
c
t
i
on
m
odel
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
V
ent
r
i
c
ul
ar
T
ac
hy
ar
r
hy
t
hm
i
a O
ns
et
P
r
e
d
ic
t
io
n
B
a
s
ed
o
n H
R
V
an
d G
en
et
i
c
…
(
K.
H
.
Bo
o
n
)
1007
R
ef
er
en
ces
[1
]
de
Lun
a
A
B
,
C
ou
m
e
l
P
,
Le
c
l
er
c
q
J
F
.
A
m
bul
a
t
or
y
s
u
dden
c
ar
di
ac
deat
h:
M
ec
hani
s
m
s
o
f
pr
oduc
t
i
on
of
f
at
al
ar
r
hy
t
hm
i
a
on t
h
e ba
s
i
s
o
f
dat
a
f
r
om
15
7 c
a
s
e
s
.
A
m
er
i
c
an
H
ear
t
J
o
ur
nal
.
19
89;
117(
1)
:
151
-
15
9.
[2
]
H
onk
o
l
a J
,
H
o
ok
a
na E
,
M
al
i
nen S
,
K
ai
k
k
one
n K
S
,
J
un
t
t
i
l
a M
J
,
I
s
ohan
ni
M
,
et
al
.
P
s
y
c
hot
r
op
i
c
m
edi
c
at
i
on
s
a
nd t
he r
i
s
k
o
f
s
udden
c
ar
di
a
c
deat
h d
ur
i
ng
a
n ac
ut
e
c
or
onar
y
ev
en
t
.
E
ur
o
pean hear
t
j
our
na
l
.
2
012;
3
3
(6
):
7
45
-
7
51
.
[3
]
T
er
es
hc
h
enk
o LG
,
F
et
i
c
s
B
J
,
D
om
i
t
r
ov
i
c
h P
P
,
Li
nds
ay
B
D
,
B
er
ger
R
D
.
P
r
e
di
c
t
i
o
n of
v
ent
r
i
c
u
l
a
r
ta
c
hy
ar
r
hy
t
h
m
i
a
s
by
i
nt
r
ac
ar
di
ac
r
epo
l
ar
i
z
at
i
on v
ar
i
a
bi
l
i
t
y
anal
y
s
i
s
.
C
i
r
c
ul
at
i
o
n A
r
r
hy
t
hm
i
a an
d
el
ec
t
r
oph
y
s
i
ol
ogy
.
20
09;
2
(3
):
276
-
2
84
.
[4
]
Z
ok
a
ee S
,
I
s
l
am
i
c
A
z
ad
U
ni
v
e
r
s
i
t
y
Q
az
v
i
n B
,
F
a
ez
K
,
A
m
i
r
k
abi
r
U
n
i
v
er
s
i
t
y
o
f
T
.
H
um
a
n I
d
ent
i
f
i
c
a
t
i
o
n
B
as
ed
on
E
C
G
an
d P
a
l
m
p
r
i
n
t.
I
nt
er
nat
i
ona
l
J
our
nal
of
E
l
e
c
t
r
i
c
al
a
nd C
om
p
ut
er
E
ngi
neer
i
n
g (
I
J
E
C
E
)
.
2012;
2
(2
):
261
-
26
6.
[5
]
B
us
on
o P
,
K
ha
s
ana
h Y
N
.
D
ev
el
opm
ent
o
f
E
C
G
F
eat
ur
e
E
x
t
r
ac
t
i
on S
of
t
w
ar
e
.
P
r
o
c
eed
i
n
g
of
t
he
E
l
ec
t
r
i
c
a
l
E
ng
i
ne
er
i
n
g C
om
put
er
S
c
i
enc
e an
d I
nf
or
m
at
i
c
s
.
20
15;
2
(1
).
[6
]
C
hi
ang
E
,
V
el
l
ai
s
am
y
M
.
I
nt
el
l
i
ge
nt
P
i
l
l
ow
f
or
H
ear
t
R
at
e
M
oni
t
or
.
I
nt
er
n
at
i
o
nal
J
our
na
l
of
E
l
ec
t
r
i
c
al
and C
om
put
er
E
ng
i
ne
er
i
n
g (
I
J
E
C
E
)
.
2013;
3
(6
):
791
-
79
6.
[7
]
R
aj
endr
a A
c
har
y
a U
,
P
aul
J
o
s
eph
K
,
K
a
nnat
hal
N
,
Li
m
C
M
,
S
ur
i
J
S
.
H
e
ar
t
r
a
t
e v
ar
i
abi
l
i
t
y
:
a r
ev
i
ew
.
M
edi
c
al
an
d B
i
o
l
og
i
c
a
l
E
ng
i
ne
er
i
ng
and C
om
put
i
n
g
.
2
006;
4
4(
12)
:
1031
-
10
51
.
[8
]
R
eed M
J
,
R
ober
t
s
on C
E
,
A
d
di
s
o
n P
S
.
H
e
ar
t
r
at
e v
ar
i
a
bi
l
i
t
y
m
eas
ur
e
m
ent
s
a
nd t
h
e pr
edi
c
t
i
on
of
v
ent
r
i
c
u
l
ar
ar
r
hy
t
hm
i
as
.
Q
J
M
:
m
ont
hl
y
j
our
nal
of
t
he A
s
s
o
c
i
a
t
i
on o
f
P
hy
s
i
c
i
a
ns
.
200
5;
9
8
(2
):
87
-
95.
[9
]
W
a
t
an
abe
M
A
.
H
ear
t
r
at
e
t
ur
b
ul
en
c
e
s
l
op
e
r
ed
uc
t
i
on
i
n
i
m
m
i
nent
v
e
nt
r
i
c
ul
ar
t
ac
hy
ar
r
hy
t
h
m
i
a
and
i
t
s
i
m
pl
i
c
at
i
o
ns
.
J
o
ur
nal
of
c
ar
d
i
o
v
as
c
ul
ar
el
ec
t
r
oph
y
s
i
ol
ogy
.
20
06;
17(
7)
:
73
5
-
7
40.
[1
0
]
S
k
i
nn
er
J
E
,
P
r
a
t
t
C
M
,
V
y
bi
r
al
T
.
A
r
e
duc
t
i
o
n i
n t
h
e c
or
r
el
at
i
on
di
m
en
s
i
on
of
he
ar
t
be
at
i
n
t
er
v
al
s
pr
ec
e
des
i
m
m
i
nent
v
ent
r
i
c
ul
ar
f
i
br
i
l
l
a
t
i
on
i
n
hu
m
an
s
u
bj
e
c
t
s
.
A
m
er
i
c
a
n H
ear
t
J
our
na
l
.
19
93
;
12
5(
3)
:
731
-
7
43
.
[1
1
]
J
oo S
,
C
ho
i
K
-
J
,
H
uh
S
-
J
.
P
r
edi
c
t
i
on
of
s
pont
aneo
u
s
v
e
nt
r
i
c
ul
ar
t
ac
hy
ar
r
hy
t
hm
i
a by
an ar
t
i
f
i
c
i
a
l
neur
al
ne
t
w
or
k
us
i
ng
par
am
e
t
er
s
g
l
ea
ned
f
r
o
m
s
hor
t
-
t
er
m
h
ear
t
r
a
t
e v
ar
i
abi
l
i
t
y
.
E
x
per
t
S
y
s
t
em
s
w
i
t
h
A
ppl
i
c
at
i
on
s
.
20
12;
39(
3)
:
386
2
-
386
6.
[1
2
]
T
hong T
,
R
ai
t
t
M
H
.
P
r
edi
c
t
i
n
g
I
m
m
i
nen
t
E
pi
s
od
es
o
f
V
ent
r
i
c
ul
ar
T
ac
hy
ar
r
hy
t
h
m
i
a U
s
i
n
g H
ear
t
R
at
e
.
P
ac
i
n
g an
d C
l
i
ni
c
al
E
l
ec
t
r
oph
y
s
i
ol
og
y
.
20
07;
30(
7)
:
874
-
8
84.
[1
3
]
R
oz
en G
,
K
ob
o R
,
B
ei
nar
t
R
,
F
el
d
m
an
S
,
S
a
punar
M
,
Lur
i
a
D
,
et
al
.
M
ul
t
i
pol
e an
al
y
s
i
s
of
hear
t
r
at
e
v
ar
i
abi
l
i
t
y
as
a
pr
edi
c
t
or
of
i
m
m
i
nent
v
ent
r
i
c
u
l
ar
ar
r
hy
t
h
m
i
as
i
n
I
C
D
p
a
ti
e
n
ts
.
P
ac
i
ng
and
c
l
i
ni
c
a
l
el
ec
t
r
oph
y
s
i
ol
ogy
:
PAC
E
.
2
013
;
36(
1
1)
:
1
342
-
134
7.
[1
4
]
W
o
l
l
m
an
n C
G
,
G
r
adau
s
R
,
B
öc
k
er
D
,
F
et
s
c
h T
,
H
i
nt
r
i
nger
F
,
H
oh G
,
et
al
.
V
ar
i
at
i
on
s
of
hear
t
r
a
t
e
v
ar
i
abi
l
i
t
y
par
a
m
et
er
s
pr
i
or
t
o
t
he
on
s
et
of
v
ent
r
i
c
ul
ar
t
ac
hy
ar
r
hy
t
hm
i
a
an
d
s
i
nu
s
t
a
c
hy
c
ar
di
a
i
n
I
C
D
pat
i
e
nt
s
.
R
es
u
l
t
s
f
r
o
m
t
he
h
ear
t
r
at
e v
ar
i
ab
i
l
i
t
y
anal
y
s
i
s
w
i
t
h aut
om
at
ed
I
C
D
s
(
H
A
W
A
I
)
r
egi
s
t
r
y
.
P
hy
s
i
ol
o
gi
c
al
M
eas
ur
em
ent
.
2
015;
3
6
(5
):
104
7.
[1
5
]
H
uang C
L,
W
a
n
g C
J
.
A
G
A
-
b
as
ed
f
eat
ur
e
s
el
ec
t
i
on
and
pa
r
am
et
er
s
opt
i
m
i
z
at
i
onf
or
s
upp
or
t
v
ec
t
o
r
m
ac
hi
ne
s
.
E
x
per
t
S
y
s
t
em
s
w
i
t
h A
ppl
i
c
a
t
i
o
ns
.
2006
;
31(
2)
:
231
-
2
40.
[1
6
]
G
ol
dber
g
er
A
L,
A
m
a
r
al
LA
N
,
G
l
as
s
L,
H
aus
dor
f
f
J
M
,
I
v
anov
P
C
,
M
ar
k
R
G
,
et
al
.
P
hy
s
i
oB
an
k
,
P
hy
s
i
oT
ool
k
i
t
,
an
d P
hy
s
i
oN
et
:
C
om
pon
ent
s
of
a N
ew
R
es
ear
c
h R
es
our
c
e f
or
C
om
p
l
ex
P
hy
s
i
ol
ogi
c
S
i
gnal
s
.
C
ir
c
ul
a
t
io
n
.
2000
;
10
1
(
23)
:
21
5
-
2
20
.
[1
7
]
C
l
i
f
f
or
d
G
D
,
T
ar
as
s
en
k
o
L.
Q
uant
i
f
y
i
ng
er
r
or
s
i
n
s
pe
c
t
r
al
e
s
t
i
m
at
es
of
H
R
V
du
e t
o b
eat
r
e
pl
ac
em
e
nt
and r
e
s
a
m
pl
i
ng.
I
E
E
E
T
r
an
s
a
c
t
i
on
s
on B
i
om
edi
c
a
l
E
ng
i
nee
r
i
ng
.
20
05;
52(
4)
:
630
-
63
8.
[1
8
]
M
c
N
am
es
J
,
T
hong
T
,
A
boy
M
.
I
m
pul
s
e
r
ej
e
c
t
i
o
n
f
i
l
t
er
f
or
ar
t
i
f
ac
t
r
em
ov
al
i
n
s
pe
c
t
r
a
l
anal
y
s
i
s
of
bi
om
edi
c
al
s
i
gna
l
s
.
E
ng
i
neer
i
n
g i
n M
edi
c
i
ne
and
B
i
o
l
ogy
S
o
c
i
et
y
,
A
nn
ual
I
nt
er
nat
i
ona
l
C
onf
er
enc
e o
f
t
h
e
I
EEE.
2004
.
[1
9
]
R
aj
endr
a A
c
har
y
a U
,
P
aul
J
o
s
eph
K
,
K
a
nnat
hal
N
,
Li
m
C
M
,
S
ur
i
J
S
.
H
e
ar
t
r
a
t
e v
ar
i
abi
l
i
t
y
:
a r
ev
i
ew
.
M
edi
c
al
&
bi
ol
o
gi
c
al
engi
neer
i
ng &
c
om
put
in
g
.
2
006;
44(
1
2)
:
1031
-
10
5
1.
[2
0
]
E
l
ec
t
r
ophy
s
i
ol
o
gy
T
F
ot
E
S
oC
t
N
A
S
oP
.
H
ear
t
R
at
e V
ar
i
abi
l
i
t
y
:
S
t
andar
ds
of
M
eas
ur
e
m
ent
,
P
hy
s
i
ol
o
gi
c
al
I
nt
er
pr
et
a
t
i
on
,
a
nd C
l
i
ni
c
al
U
s
e.
C
i
r
c
u
la
t
i
on
.
199
6
;
93(
5)
:
1
043
-
10
6
5.
[2
1
]
H
i
c
k
ey
B
,
H
eneg
han C
.
S
c
r
e
e
ni
ng
f
or
p
ar
ox
y
s
m
a
l
at
r
i
al
f
i
b
r
i
l
l
at
i
on
us
i
ng
at
r
i
a
l
pr
em
at
ur
e
c
o
nt
r
ac
t
i
on
s
and s
pe
c
t
r
al
m
eas
ur
es
.
C
om
p
ut
er
s
i
n C
ar
di
o
l
ogy
.
2
002
.
[2
2
]
A
ni
t
a B
,
F
er
nan
do S
oar
es
S
,
A
na P
aul
a R
,
A
r
gent
i
na
L.
A
s
t
udy
on t
he op
t
i
m
u
m
or
der
o
f
aut
or
e
gr
es
s
i
v
e
m
od
el
s
f
or
h
ea
r
t
r
at
e v
ar
i
abi
l
i
t
y
.
P
h
y
s
i
ol
o
gi
c
al
M
eas
u
r
em
ent
.
20
02;
23(
2)
:
3
25.
[2
3
]
P
adm
av
at
hi
K
,
G
ok
ar
aj
u
R
a
ngar
aj
u I
ns
t
i
t
ut
e
of
E
,
T
ec
h
nol
ogy
,
R
am
a
k
r
i
s
hn
a K
S
,
V
el
aga
pud
i
R
am
ak
r
i
s
hna S
i
d
dhar
t
ha E
ngi
neer
i
n
g C
.
C
l
as
s
i
f
i
c
at
i
on of
E
C
G
s
i
gnal
dur
i
ng A
t
r
i
a
l
F
i
br
i
l
l
at
i
on u
s
i
n
g
B
ur
g’
s
m
et
ho
d.
I
nt
er
nat
i
ona
l
J
our
nal
of
E
l
ec
t
r
i
c
a
l
and
C
om
p
ut
er
E
ngi
n
eer
i
ng
(
I
J
E
C
E
)
.
2
015
;
5
(1
):
64
-
70.
[2
4
]
Sp
y
e
rs
-
A
s
h
by
J
M
,
B
ai
n P
G
,
R
ober
t
s
S
J
.
A
c
om
par
i
s
on
of
f
as
t
f
o
ur
i
er
t
r
a
ns
f
or
m
(
F
F
T
)
and
aut
or
e
gr
es
s
i
v
e (
A
R
)
s
p
ec
t
r
a
l
es
t
i
m
a
t
i
on
t
e
c
hn
i
que
s
f
or
t
he an
al
y
s
i
s
o
f
t
r
e
m
or
dat
a.
J
our
nal
of
N
eur
os
c
i
e
nc
e
M
et
hods
.
1
998;
83(
1)
:
35
-
4
3.
[2
5
]
N
i
k
i
a
s
C
L,
R
a
ghuv
eer
M
R
.
B
i
s
p
ec
t
r
um
e
s
t
i
m
at
i
on:
A
di
gi
t
al
s
i
g
nal
pr
o
c
es
s
i
n
g f
r
am
ew
or
k
.
P
r
oc
ee
di
n
gs
o
f
t
h
e I
E
E
E
.
19
87
;
7
5
(7
):
8
69
-
8
91
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
99
–
1
0
08
1008
[2
6
]
P
i
nha
s
I
,
T
ol
ed
o E
,
A
r
av
ot
D
,
A
k
s
e
l
r
od S
.
B
i
c
o
her
en
c
e
anal
y
s
i
s
of
new
c
ar
di
ov
as
c
ul
ar
s
pe
c
t
r
a
l
c
om
pone
nt
s
o
bs
er
v
ed i
n he
ar
t
-
t
r
ans
pl
a
nt
pat
i
ent
s
:
s
t
at
i
s
t
i
c
al
appr
oa
c
h f
or
b
i
c
o
her
e
nc
e t
hr
es
ho
l
di
ng.
I
E
E
E
T
r
an
s
ac
t
i
on
s
o
n B
i
om
ed
i
c
al
E
ngi
n
eer
i
ng
.
2004
;
51(
10)
:
17
74
-
17
8
3.
[2
7
]
Y
u S
N
,
Lee M
Y
.
B
i
s
pec
t
r
a
l
a
nal
y
s
i
s
an
d ge
net
i
c
a
l
gor
i
t
hm
f
or
c
o
nge
s
t
i
v
e h
ear
t
f
ai
l
ur
e r
ec
og
ni
t
i
o
n
bas
e
d on
hear
t
r
at
e
v
ar
i
ab
i
l
i
t
y
.
C
om
put
er
s
i
n bi
o
l
og
y
an
d
m
e
di
c
i
ne
.
2012
;
42(
8)
:
81
6
-
8
25
.
[2
8
]
Z
hou
S
M
,
G
an
J
Q
,
S
epu
l
v
eda
F
.
C
l
a
s
s
i
f
y
i
ng
m
ent
a
l
t
a
s
k
s
ba
s
e
d on f
eat
ur
es
of
hi
g
her
-
or
de
r
s
t
at
i
s
t
i
c
s
f
r
om
E
E
G
s
i
gn
al
s
i
n
br
ai
n
–
c
o
m
put
er
i
n
t
er
f
a
c
e.
I
nf
or
m
at
i
on S
c
i
en
c
e
s
.
2
008;
178(
6)
:
1
629
-
16
40.
[2
9
]
P
i
nc
u
s
S
M
,
G
ol
dber
ger
A
L.
P
hy
s
i
ol
o
gi
c
al
t
i
m
e
-
s
er
i
es
a
nal
y
s
i
s
:
w
hat
doe
s
r
egu
l
ar
i
t
y
quant
i
f
y
?
A
m
er
i
c
an
J
our
nal
of
P
h
y
s
i
ol
og
y
-
H
ear
t
and
C
i
r
c
ul
at
or
y
P
hy
s
i
ol
og
y
.
1
994;
266(
4)
:
1
643
-
16
56
.
[3
0
]
M
ohebbi
M
,
G
has
s
e
m
i
a
n H
.
P
r
edi
c
t
i
on
of
par
ox
y
s
m
al
at
r
i
a
l
f
i
br
i
l
l
a
t
i
on
bas
ed on
non
-
l
i
n
e
ar
anal
y
s
i
s
and s
p
ec
t
r
um
an
d bi
s
pe
c
t
r
u
m
f
eat
ur
es
of
t
he he
ar
t
r
at
e
v
ar
i
abi
l
i
t
y
s
i
g
nal
.
C
om
put
er
m
et
hod
s
an
d
pr
ogr
am
s
i
n
b
i
om
edi
c
i
ne
.
20
12
;
105(
1)
:
40
-
4
9.
[3
1
]
C
or
t
es
C
,
V
apn
i
k
V
.
S
upp
or
t
-
V
ec
t
or
N
et
w
or
k
s
.
M
ac
hi
n
e Le
ar
ni
ng
.
20(
3)
:
2
73
-
2
9
7.
[3
2
]
N
ar
i
n A
,
I
s
l
er
Y
,
O
z
er
M
.
I
nv
es
t
i
g
at
i
n
g t
he
per
f
or
m
an
c
e
i
m
p
r
ov
em
ent
of
H
R
V
I
nd
i
c
es
i
n
C
H
F
us
i
ng
f
eat
ur
e
s
el
ec
t
i
on
m
et
ho
ds
ba
s
ed
on
ba
c
k
w
ar
d
el
i
m
i
n
at
i
o
n
a
nd
s
t
at
i
s
t
i
c
a
l
s
i
gni
f
i
c
anc
e.
C
o
m
put
er
s
i
n
bi
ol
ogy
and
m
e
di
c
i
ne
.
201
4;
45
:
72
-
7
9.
[3
3
]
Y
u
w
ono T
,
S
et
i
aw
an N
A
,
N
ugr
oho H
A
,
P
er
s
ada A
G
,
P
r
as
oj
o I
,
D
ew
i
S
K
,
et
al
.
D
ec
i
s
i
o
n S
uppor
t
S
y
s
t
em
f
o
r
H
ear
t
D
i
s
e
as
e
D
i
agno
s
i
ng
U
s
i
ng
K
-
N
N
A
l
gor
i
t
hm
.
P
r
oc
eedi
ng
o
f
t
h
e
E
l
ec
t
r
i
c
a
l
E
ngi
ne
er
i
n
g C
om
put
e
r
S
c
i
en
c
e and
I
nf
or
m
at
i
c
s
.
2
015;
2
(1
).
[3
4
]
K
i
l
hw
an K
,
U
ns
un C
,
Y
unho J
,
J
aes
eok
K
,
e
d
i
to
r
s
.
D
es
i
gn a
nd i
m
pl
em
ent
at
i
o
n of
bi
om
edi
c
al
S
oC
f
o
r
i
m
pl
ant
a
bl
e
c
ar
di
o
v
er
t
er
def
i
br
i
l
l
at
or
s
. S
o
l
i
d
-
S
t
at
e C
i
r
c
ui
t
s
C
o
n
f
e
re
n
c
e
,
ASSC
C
'
0
7
I
EEE As
i
a
n
.
2007
.
[3
5
]
M
as
s
agr
am
W
,
H
af
n
er
N
,
C
hen
M
,
M
ac
c
hi
ar
u
l
o L,
Lu
bec
k
e
V
M
,
B
or
i
c
-
Lub
ec
k
e O
.
D
i
gi
t
a
l
H
ear
t
-
R
a
te
Va
ri
a
b
i
l
i
t
y
P
ar
a
m
et
er
M
oni
t
or
i
ng
and
A
s
s
e
s
s
m
en
t
A
S
I
C
.
I
E
E
E
T
r
ans
a
c
t
i
ons
on
B
i
om
ed
i
c
al
C
i
r
c
u
i
t
s
and S
y
s
t
em
s
.
20
10;
4
(1
):
19
-
26
.
[3
6
]
F
ang
W
C
,
H
uang
H
C
,
T
s
eng
S
Y
.
D
es
i
gn of
hear
t
r
at
e v
ar
i
a
bi
l
i
t
y
pr
oc
e
s
s
or
f
or
por
t
a
bl
e
3
-
l
ead E
C
G
m
oni
t
or
i
ng s
y
s
t
em
-
on
-
c
h
i
p.
Ex
p
e
rt
Sy
s
t
em
s
w
i
t
h A
ppl
i
c
at
i
on
s
.
2013
;
40(
5)
:
1
491
-
1
504.
Evaluation Warning : The document was created with Spire.PDF for Python.