I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
3
9
,
No
.
2
,
A
u
g
u
s
t
2
0
2
5
,
p
p
.
1
164
~
1
1
7
5
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
3
9
.i
2
.
pp
1
1
6
4
-
1
1
7
5
1164
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Adv
a
nced
dee
p at
tent
io
n
neura
l in
f
erence
network
f
o
r
enha
nced arrhy
t
h
mia
det
ec
tion a
nd
accura
te
cla
ss
ific
a
tion
H
.
Su
m
it
ha
,
M
.
Dev
a
na
t
ha
n
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
a
n
d
C
o
mp
u
t
e
r
En
g
i
n
e
e
r
i
n
g
,
R
e
v
a
U
n
i
v
e
r
si
t
y
,
B
a
n
g
a
l
o
r
e
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Au
g
17
,
2
0
2
4
R
ev
is
ed
Ap
r
17
,
2
0
2
5
Acc
ep
ted
J
u
l
2
,
2
0
2
5
Arrh
y
t
h
m
ias
a
re
irreg
u
lar
h
e
a
rtb
e
a
ts
th
a
t
c
a
n
lea
d
to
se
v
e
re
h
e
a
lt
h
risk
s,
in
c
lu
d
in
g
s
u
d
d
e
n
c
a
rd
iac
d
e
a
th
,
n
e
c
e
ss
it
a
ti
n
g
a
c
c
u
ra
te
a
n
d
t
ime
ly
d
e
tec
ti
o
n
fo
r
e
ffe
c
ti
v
e
trea
tme
n
t.
Trad
it
io
n
a
l
d
iag
n
o
stic
m
e
th
o
d
s
su
c
h
a
s
stre
ss
tes
ts,
re
stin
g
e
lec
tro
c
a
rd
i
o
g
ra
m
s
(ECG
s
),
a
n
d
2
4
-
h
o
u
r
Ho
lt
e
r
m
o
n
it
o
rs
a
re
li
m
it
e
d
b
y
t
h
e
ir
m
o
n
i
to
ri
n
g
c
a
p
a
c
it
y
a
n
d
o
fte
n
re
su
lt
i
n
d
e
lay
e
d
d
iag
n
o
se
s,
c
o
m
p
ro
m
isin
g
p
a
ti
e
n
t
sa
fe
ty
.
To
a
d
d
re
ss
th
e
se
c
h
a
ll
e
n
g
e
s,
t
h
is
p
a
p
e
r
in
tro
d
u
c
e
s
th
e
d
e
e
p
a
tt
e
n
ti
o
n
n
e
u
ra
l
in
fe
re
n
c
e
n
e
two
r
k
(DA
NIN
)
m
e
th
o
d
o
lo
g
y
.
DA
NIN
i
n
teg
ra
tes
o
n
e
-
d
ime
n
sio
n
a
l
ECG
sig
n
a
ls
with
two
-
d
ime
n
sio
n
a
l
s
p
e
c
tral
ima
g
e
s
u
sin
g
m
u
lt
i
-
m
o
d
a
l
fe
a
tu
re
fu
si
o
n
,
c
a
p
tu
rin
g
c
o
m
p
re
h
e
n
s
iv
e
c
a
rd
iac
in
fo
rm
a
ti
o
n
i
n
b
o
th
tem
p
o
ra
l
a
n
d
fre
q
u
e
n
c
y
d
o
m
a
in
s.
Th
e
m
e
th
o
d
o
lo
g
y
e
m
p
l
o
y
s
a
d
v
a
n
c
e
d
d
e
e
p
a
tt
e
n
ti
o
n
n
e
two
r
k
-
b
a
s
e
d
m
o
d
e
ls
fo
r
s
u
p
e
rio
r
fe
a
tu
re
e
x
trac
ti
o
n
,
re
c
o
g
n
izi
n
g
i
n
tri
c
a
te
p
a
tt
e
r
n
s
a
n
d
l
o
n
g
-
ra
n
g
e
d
e
p
e
n
d
e
n
c
ies
with
i
n
th
e
d
a
ta.
Ad
d
it
i
o
n
a
l
ly
,
t
h
e
i
n
c
lu
sio
n
o
f
a
n
in
fe
re
n
c
e
m
o
d
e
l
sy
ste
m
e
n
h
a
n
c
e
s
in
terp
re
tab
il
it
y
a
n
d
u
sa
b
il
i
ty
,
m
a
k
in
g
t
h
e
m
o
d
e
l
h
ig
h
ly
su
it
a
b
le.
F
u
rth
e
r,
DA
NI
N
is
e
v
a
lu
a
ted
c
o
n
sid
e
ri
n
g
t
h
e
M
IT
-
BIH
d
a
tas
e
t,
a
n
d
e
x
ten
si
v
e
c
o
m
p
a
ra
ti
v
e
a
n
a
ly
sis
wit
h
sta
te
-
of
-
th
e
-
a
rt
t
e
c
h
n
iq
u
e
s
d
e
m
o
n
stra
tes
th
a
t
DA
NIN
sig
n
ifi
c
a
n
tl
y
imp
ro
v
e
s
a
c
c
u
ra
c
y
,
p
re
c
isio
n
,
re
c
a
ll
,
a
n
d
F1
-
sc
o
re
,
h
ig
h
li
g
h
ti
n
g
i
ts
p
o
t
e
n
ti
a
l
to
re
v
o
lu
ti
o
n
ize
a
rrh
y
th
m
ia
d
e
tec
ti
o
n
a
n
d
imp
r
o
v
e
p
a
ti
e
n
t
o
u
tco
m
e
s.
K
ey
w
o
r
d
s
:
Ar
r
h
y
th
m
ia
d
etec
tio
n
Atten
tio
n
n
etwo
r
k
Dee
p
lear
n
in
g
E
C
G
s
ig
n
al
an
aly
s
is
Featu
r
e
ex
tr
ac
tio
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
H
.
Su
m
ith
a
Dep
ar
tm
en
t o
f
E
lectr
ical
an
d
C
o
m
p
u
ter
E
n
g
in
ee
r
in
g
,
R
ev
a
Un
iv
er
s
ity
B
an
g
alo
r
e
,
I
n
d
ia
E
m
ail: su
m
ith
ah
_
1
2
@
r
ed
if
f
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Ar
r
h
y
th
m
ias
a
r
e
a
s
ig
n
if
ican
t
class
o
f
ca
r
d
iac
d
is
ea
s
es
th
at
r
eq
u
ir
e
p
r
o
p
er
d
iag
n
o
s
is
.
T
h
is
is
b
ec
au
s
e
ar
r
h
y
th
m
ias
r
aise
th
e
p
o
s
s
ib
ilit
y
o
f
h
ig
h
-
r
is
k
i
n
cid
en
ts
,
s
u
ch
as
u
n
e
x
p
ec
ted
ca
r
d
iac
d
ea
th
.
An
ac
cu
r
ate
d
iag
n
o
s
is
m
u
s
t b
e
m
ad
e
q
u
ick
ly
to
tr
ea
t th
e
ch
an
g
es in
th
e
e
lectr
o
ca
r
d
io
g
r
am
(
E
C
G)
wh
ich
h
av
e
p
r
o
v
en
to
b
e
d
if
f
icu
lt
to
c
o
r
r
ec
tly
d
etec
t
u
s
in
g
tr
ad
itio
n
al
a
u
to
m
ated
ap
p
r
o
ac
h
es
f
o
r
ar
r
h
y
th
m
ia
d
etec
tio
n
.
Stan
d
ar
d
d
iag
n
o
s
tic
tech
n
iq
u
es in
clu
d
in
g
s
tr
ess
test
s
,
r
esti
n
g
E
C
Gs,
a
n
d
2
4
-
h
o
u
r
Ho
lter
m
o
n
ito
r
s
ar
e
f
r
eq
u
e
n
tly
u
s
ed
to
id
en
tify
ar
r
h
y
th
m
ias
[
1
]
.
T
h
e
lim
ited
ca
r
d
iac
r
ec
o
r
d
in
g
ca
p
ac
ity
,
wh
ich
allo
ws
ab
o
u
t
2
,
0
0
0
h
ea
r
tb
ea
ts
to
b
e
r
ec
o
r
d
e
d
d
u
r
in
g
a
s
tr
ess
te
s
t,
s
till
r
es
tr
ic
ts
th
e
m
o
n
ito
r
in
g
ca
p
ab
ilit
ies
o
f
th
e
test
s
.
T
h
e
p
atien
ts
m
u
s
t
en
d
u
r
e
len
g
th
y
wait
tim
es
f
o
r
ex
am
in
atio
n
r
esu
lts
,
d
o
cto
r
ap
p
o
i
n
tm
en
ts
,
an
d
h
o
s
p
ital
s
tay
s
,
it
m
i
g
h
t
tak
e
u
p
to
f
o
u
r
m
o
n
th
s
to
g
et
test
r
esu
lts
.
T
h
e
co
n
v
e
n
tio
n
al
m
et
h
o
d
s
th
at
a
r
e
b
ein
g
p
r
o
p
o
s
ed
in
v
o
lv
e
m
a
k
in
g
a
p
p
o
in
tm
e
n
ts
with
m
ed
ical
d
o
cto
r
s
in
p
er
s
o
n
an
d
p
er
f
o
r
m
i
n
g
a
s
er
ies
o
f
d
iag
n
o
s
tic
test
s
.
T
h
e
p
atien
t
’
s
h
ea
lth
m
ay
b
e
af
f
ec
ted
b
y
th
ese
cir
cu
m
s
tan
c
es
s
in
ce
th
er
e
i
s
a
d
ec
r
ea
s
ed
l
ik
elih
o
o
d
o
f
ea
r
l
y
d
etec
tio
n
.
T
h
e
m
o
s
t
co
m
m
o
n
m
eth
o
d
u
s
ed
to
d
i
ag
n
o
s
e
ar
r
h
y
th
m
ias
is
an
E
C
G
r
ec
o
r
d
in
g
.
T
o
r
ec
o
r
d
an
d
a
n
aly
ze
th
e
elec
tr
ical
ac
tiv
ity
o
f
th
e
h
ea
r
t,
elec
tr
o
d
es
ca
n
b
e
p
l
ac
ed
o
n
th
e
s
k
in
an
d
lef
t
th
er
e
f
o
r
a
p
r
ed
ef
in
e
d
p
er
io
d
.
T
h
e
elec
tr
ical
ac
tiv
ity
o
f
th
e
h
ea
r
t
m
ay
b
e
ass
ess
ed
f
r
o
m
a
v
ar
iety
o
f
a
n
g
les
a
n
d
o
r
i
en
tatio
n
s
u
s
in
g
E
C
Gs.
As
s
u
ch
,
th
ey
ca
n
ac
t
as
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
d
va
n
ce
d
d
ee
p
a
tten
tio
n
n
eu
r
a
l in
feren
ce
n
etw
o
r
k
fo
r
en
h
a
n
ce
d
a
r
r
h
yth
mia
…
(
H.
S
u
mit
h
a
)
1165
in
d
icato
r
s
f
o
r
a
r
an
g
e
o
f
d
is
ea
s
es.
T
h
e
p
r
im
ar
y
r
esp
o
n
s
ib
ilit
y
o
f
th
e
E
C
G
lead
s
is
t
o
d
etec
t
an
d
ass
es
s
an
o
m
alies in
wav
ef
o
r
m
s
an
d
r
h
y
th
m
s
.
An
E
C
G
is
a
co
m
p
r
e
h
en
s
iv
e
r
ec
o
r
d
o
f
th
e
elec
tr
ica
l c
h
ar
ac
ter
is
t
ics o
f
a
h
ea
r
tb
ea
t.
E
ac
h
o
f
th
e
s
ev
er
al
wav
ef
o
r
m
s
th
at
c
o
m
p
r
is
e
an
E
C
G
s
ig
n
al
r
ep
r
esen
ts
a
s
in
g
le
p
u
ls
e.
Ov
er
th
e
p
ast
s
ev
er
al
y
ea
r
s
,
th
er
e
h
as
b
ee
n
a
s
izab
le
s
u
r
g
e
in
th
e
u
s
a
g
e
o
f
E
C
G
m
o
n
ito
r
in
g
tec
h
n
o
lo
g
y
in
th
e
m
ed
ical
in
d
u
s
tr
y
.
T
h
e
o
b
s
er
v
e
d
p
atte
r
n
ca
n
b
e
attr
ib
u
ted
to
a
d
v
a
n
ce
m
en
ts
in
e
n
ab
lin
g
tech
n
o
lo
g
ies,
wh
ich
h
av
e
g
r
ea
tly
in
cr
ea
s
ed
th
e
ca
p
ab
i
liti
es
o
f
th
ese
s
y
s
tem
s
.
Fo
r
th
ese
s
y
s
tem
s
to
f
u
n
ctio
n
co
r
r
ec
tly
,
s
ev
er
al
tech
n
o
lo
g
ies
m
u
s
t
b
e
in
teg
r
at
ed
,
in
clu
d
in
g
ed
g
e
co
m
p
u
tin
g
,
m
o
b
ile
co
m
p
u
tin
g
,
an
d
th
e
i
n
ter
n
et
o
f
th
in
g
s
.
T
h
ese
tech
n
o
lo
g
ies
h
a
v
e
s
ev
e
r
al
o
th
er
u
s
es
in
ad
d
itio
n
to
b
ein
g
em
p
lo
y
ed
in
th
e
d
iag
n
o
s
is
an
d
tr
ea
tm
en
t
o
f
h
ea
lth
p
r
o
b
lem
s
.
T
h
e
d
ev
ice
s
atis
f
ie
s
s
p
ec
if
ic
m
o
d
e
-
r
elate
d
n
ee
d
s
,
tr
ac
k
s
ac
tiv
ities
,
an
d
en
h
a
n
ce
s
s
p
o
r
ts
p
er
f
o
r
m
an
ce
[
2
]
.
Nu
m
er
o
u
s
s
tu
d
ies
h
av
e
b
ee
n
c
o
n
d
u
cted
o
n
v
ar
io
u
s
tech
n
i
q
u
es
f
o
r
u
s
in
g
d
ee
p
lear
n
in
g
(
DL
)
an
d
m
ac
h
in
e
lear
n
in
g
(
ML
)
alg
o
r
ith
m
s
to
class
if
y
ca
r
d
iac
ar
r
h
y
th
m
ias
also
v
ar
io
u
s
tec
h
n
iq
u
es
h
a
v
e
b
ee
n
d
ev
elo
p
e
d
b
y
E
C
G
an
al
y
s
is
r
esear
ch
er
s
to
a
u
to
m
atica
lly
id
e
n
tify
h
ea
r
t
ar
r
h
y
th
m
ias.
T
h
ese
tech
n
iq
u
es
em
p
lo
y
tr
ad
itio
n
al
ML
tech
n
iq
u
es.
T
h
ese
m
eth
o
d
s
ty
p
ically
co
n
s
is
t
o
f
th
r
ee
p
r
im
ar
y
s
tag
es:
p
r
ep
r
o
ce
s
s
in
g
th
e
s
ig
n
al,
ex
tr
ac
tin
g
f
ea
tu
r
es,
an
d
r
ec
o
g
n
izin
g
an
d
ca
teg
o
r
izin
g
p
atter
n
s
.
T
h
e
p
r
o
ce
s
s
o
f
f
ea
tu
r
e
ex
t
r
ac
tio
n
h
as
a
m
ajo
r
i
m
p
ac
t
o
n
h
o
w
well
h
ea
r
tb
ea
t
ca
teg
o
r
izatio
n
wo
r
k
s
.
Am
o
n
g
th
e
o
f
ten
-
u
s
ed
f
ea
t
u
r
e
ex
tr
a
ctio
n
tech
n
iq
u
es
ar
e
p
r
in
cip
al
co
m
p
o
n
e
n
t
an
aly
s
i
s
(
PC
A)
,
wav
elet
tr
an
s
f
o
r
m
(
W
T
)
,
d
is
cr
ete
wav
elet
tr
an
s
f
o
r
m
s
(
DW
T
)
,
in
d
ep
en
d
en
t
co
m
p
o
n
e
n
t
an
aly
s
is
(
I
C
A)
,
an
d
o
th
er
m
an
u
all
y
d
ev
elo
p
e
d
f
ea
tu
r
es.
T
h
ey
e
m
p
lo
y
ed
a
PC
A
to
id
en
tify
f
ea
tu
r
es
an
d
d
ec
r
ea
s
e
th
e
d
im
en
s
io
n
ality
o
f
th
e
E
C
G
d
ata.
T
o
co
n
s
tr
u
ct
th
e
atr
ial
f
ib
r
illatio
n
(
AF)
d
etec
to
r
,
it
co
m
b
in
es
th
e
DW
T
with
m
o
r
p
h
o
l
o
g
y
to
ex
tr
ac
t
f
ea
tu
r
es.
W
h
en
c
o
m
b
in
e
d
wi
th
th
is
m
eth
o
d
,
th
e
im
p
lem
en
tatio
n
o
f
an
ar
tific
i
al
n
eu
r
al
n
etwo
r
k
(
ANN)
class
if
ier
p
r
o
d
u
ce
d
p
o
s
itiv
e
r
e
s
u
lts
.
I
t
h
as
b
ee
n
d
em
o
n
s
tr
ated
th
at
W
T
is
ef
f
ec
tiv
e
in
in
ter
p
r
etin
g
E
C
G
d
ata
d
u
e
to
th
e
s
ig
n
als
’
in
h
er
e
n
t n
o
n
-
s
tatio
n
ar
ity
.
B
y
ex
tr
ac
tin
g
ce
r
tain
ch
ar
ac
t
er
is
tics
f
r
o
m
E
C
G
ar
r
h
y
th
m
ia
s
ig
n
als,
clas
s
if
icatio
n
m
o
d
els
th
at
ca
n
d
is
tin
g
u
is
h
b
etwe
en
d
if
f
e
r
en
t
ty
p
es
o
f
a
r
r
h
y
th
m
ias
ar
e
d
e
v
elo
p
ed
.
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
s
(
SVMs)
an
d
ANNs
ar
e
th
e
two
tech
n
iq
u
es
f
o
r
h
a
n
d
lin
g
ca
teg
o
r
izatio
n
p
r
o
b
lem
s
wh
ich
h
a
v
e
p
r
esen
te
d
a
m
eth
o
d
f
o
r
t
h
e
au
to
m
atic
class
if
icatio
n
o
f
E
C
G
d
ata
u
s
in
g
m
u
ltip
le
SVMs.
E
lh
aj
et
a
l.
[
3
]
e
m
p
lo
y
e
d
a
h
y
b
r
i
d
ap
p
r
o
ac
h
u
tili
zin
g
two
SVMs
to
d
etec
t
AF
.
I
t
is
wid
ely
ac
ce
p
ted
t
h
at
th
e
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
(
ML
P)
is
th
e
m
o
s
t
o
f
ten
u
s
ed
ANN
d
esig
n
f
o
r
ca
teg
o
r
i
zin
g
ar
r
h
y
th
m
ias
.
T
h
e
s
u
b
ject
o
f
r
ec
o
g
n
izin
g
ar
r
h
y
th
m
ias h
as b
ee
n
s
tu
d
ied
ex
t
en
s
iv
ely
an
d
u
s
ed
with
DL
wit
h
ef
f
icien
t
r
esu
lts
.
B
ec
au
s
e
DL
p
er
f
o
r
m
s
well
in
s
o
m
an
y
d
if
f
er
en
t
ap
p
licatio
n
s
-
s
u
ch
as
p
h
o
to
id
en
tific
atio
n
,
s
p
ee
ch
r
ec
o
g
n
itio
n
,
an
d
m
ac
h
i
n
e
v
is
io
n
-
it
is
in
cr
ed
ib
ly
p
o
ten
t.
T
we
lv
e
d
is
tin
ct
r
h
y
th
m
p
atter
n
s
wer
e
class
if
ied
u
s
in
g
a
d
ee
p
n
eu
r
al
n
etwo
r
k
(
DN
N)
d
ev
elo
p
ed
.
T
h
e
DNN
ac
cu
r
ac
y
was
ar
b
itra
ted
s
u
f
f
icien
t
f
o
r
th
is
task
,
to
class
if
y
ar
r
h
y
th
m
ias,
wh
ich
em
p
lo
y
a
DL
m
eth
o
d
o
l
o
g
y
.
T
h
e
r
esear
ch
er
s
em
p
lo
y
ed
an
lo
n
g
s
h
o
r
t
-
t
er
m
m
em
o
r
y
n
etwo
r
k
(
L
STM
)
-
a
s
p
ec
if
ic
ty
p
e
o
f
n
e
u
r
al
n
etw
o
r
k
-
to
ac
h
iev
e
th
is
.
B
ad
r
et
a
l.
[
4
]
d
escr
ib
ed
an
au
to
m
ated
tech
n
iq
u
e
f
o
r
ca
te
g
o
r
izin
g
ca
r
d
iac
a
r
r
h
y
th
m
ias
in
a
r
esear
ch
s
tu
d
y
.
T
h
e
m
o
d
el
em
p
lo
y
e
d
a
1
D
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN)
tech
n
iq
u
e.
T
h
e
a
u
th
o
r
s
o
f
th
e
s
tu
d
y
d
ev
elo
p
e
d
a
2
D
-
C
NN
m
o
d
el
to
class
if
y
ar
r
h
y
th
m
ias.
T
h
e
s
y
s
tem
u
s
es
th
e
w
h
o
le
f
ea
tu
r
e
m
a
p
s
o
f
h
ea
r
tb
ea
ts
th
at
ar
e
g
en
e
r
ated
v
ia
em
p
ir
ical
m
o
d
al
d
ec
o
m
p
o
s
itio
n
.
Pre
v
io
u
s
r
esear
ch
em
p
l
o
y
ed
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
(
R
NNs)
to
d
is
cr
im
in
ate
b
etwe
en
ab
n
o
r
m
al
an
d
n
o
r
m
al
h
ea
r
tb
ea
ts
.
I
t
is
n
o
t
o
b
v
io
u
s
h
o
w
f
ea
tu
r
e
e
x
tr
ac
tio
n
an
d
m
o
d
u
le
class
if
ic
atio
n
v
ar
y
in
DL
s
y
s
tem
s
.
I
n
s
t
ea
d
,
th
ese
two
task
s
ar
e
in
teg
r
ated
in
to
a
s
in
g
le,
s
m
o
o
th
p
r
o
ce
s
s
.
T
h
e
DL
alg
o
r
ith
m
s
em
p
lo
y
lar
g
e
v
o
l
u
m
es o
f
E
C
G
d
ata
to
au
t
o
m
atica
lly
id
en
tif
y
t
h
e
cr
u
cial
elem
en
ts
n
ee
d
ed
f
o
r
ca
teg
o
r
izatio
n
.
E
v
en
in
ca
s
es
wh
en
o
p
er
ato
r
in
ter
ac
tio
n
is
n
o
t
r
eq
u
ir
ed
,
in
t
er
p
r
etab
ilit
y
r
e
m
ain
s
a
s
ig
n
if
i
ca
n
t
ch
allen
g
e
wh
en
u
s
in
g
DL
ap
p
r
o
ac
h
es.
On
e
b
en
ef
it
o
f
DL
s
y
s
tem
s
is
th
eir
ab
ilit
y
to
au
to
m
atica
lly
ex
tr
ac
t
ch
ar
ac
ter
is
tics
f
r
o
m
u
n
p
r
o
ce
s
s
ed
,
r
aw
d
ata.
Du
e
to
its
p
o
ten
tial
to
in
f
lict
p
ain
o
n
m
ed
ical
s
taf
f
,
th
is
is
s
u
e
is
o
f
u
tm
o
s
t
im
p
o
r
ta
n
ce
in
t
h
e
f
ield
o
f
m
ed
ical
ap
p
licatio
n
s
[
5
]
.
Scien
tis
ts
h
av
e
u
s
ed
a
v
ar
iet
y
o
f
ad
v
an
ce
d
d
ee
p
-
lear
n
in
g
alg
o
r
ith
m
s
to
c
lass
if
y
ar
r
h
y
th
m
ias
.
Ho
wev
er
,
th
er
e
h
as b
ee
n
litt
le
im
p
r
o
v
em
e
n
t in
ca
teg
o
r
izatio
n
p
er
f
o
r
m
an
ce
.
T
h
e
m
o
tiv
atio
n
f
o
r
th
is
r
esear
ch
o
n
ar
r
h
y
th
m
ia
d
etec
tio
n
an
d
class
if
icatio
n
is
d
r
iv
en
b
y
t
h
e
n
ee
d
to
en
h
an
ce
th
e
ac
c
u
r
ac
y
a
n
d
ti
m
elin
ess
o
f
d
iag
n
o
s
in
g
t
h
ese
p
o
ten
tially
f
atal
ca
r
d
iac
c
o
n
d
i
tio
n
s
.
Ar
r
h
y
th
m
ias,
wh
ich
ca
n
lead
to
s
u
d
d
en
ca
r
d
iac
d
ea
th
,
p
o
s
e
a
s
ig
n
if
ica
n
t
h
ea
lth
r
is
k
th
at
n
ec
ess
itates
p
r
o
m
p
t
an
d
p
r
ec
is
e
id
en
tific
atio
n
.
T
r
ad
itio
n
al
d
ia
g
n
o
s
tic
m
eth
o
d
s
,
in
clu
d
in
g
s
tr
ess
test
s
,
r
esti
n
g
E
C
Gs,
an
d
2
4
-
h
o
u
r
H
o
lter
m
o
n
ito
r
s
,
ar
e
o
f
ten
lim
ited
in
th
eir
m
o
n
ito
r
in
g
ca
p
ac
ity
an
d
ca
n
r
esu
lt
in
d
elay
ed
d
iag
n
o
s
es,
th
er
e
b
y
co
m
p
r
o
m
is
in
g
p
atien
t
s
af
ety
.
W
ith
th
e
ad
v
e
n
t
o
f
a
d
v
an
ce
d
E
C
G
m
o
n
ito
r
in
g
tech
n
o
lo
g
ies
an
d
th
e
in
teg
r
atio
n
o
f
cu
ttin
g
-
ed
g
e
DL
a
n
d
ML
alg
o
r
ith
m
s
,
th
e
r
e
is
an
u
n
p
r
e
ce
d
en
ted
o
p
p
o
r
tu
n
ity
to
r
ev
o
lu
tio
n
ize
ar
r
h
y
th
m
ia
d
etec
tio
n
.
is
r
esear
ch
aim
s
to
lev
er
ag
e
th
ese
tech
n
o
lo
g
ical
a
d
v
an
ce
m
e
n
ts
to
d
ev
elo
p
m
o
r
e
ac
cu
r
ate,
ef
f
icien
t,
an
d
r
eliab
le
d
ia
g
n
o
s
tic
to
o
ls
,
u
ltima
tely
im
p
r
o
v
in
g
ea
r
l
y
d
etec
tio
n
an
d
p
atien
t
o
u
tco
m
e
s
,
an
d
r
ed
u
cin
g
th
e
b
u
r
d
e
n
o
n
h
ea
lth
ca
r
e
s
y
s
tem
s
.
−
Featu
r
e
f
u
s
io
n
:
th
e
d
ee
p
atten
tio
n
n
eu
r
al
in
f
e
r
en
ce
n
etwo
r
k
(
DANI
N
)
m
eth
o
d
o
lo
g
y
in
tr
o
d
u
ce
s
a
n
o
v
el
ap
p
r
o
ac
h
b
y
in
teg
r
atin
g
o
n
e
-
d
im
en
s
io
n
al
E
C
G
d
ata
with
t
w
o
-
d
im
en
s
io
n
al
s
p
ec
tr
al
im
ag
es
th
r
o
u
g
h
m
u
lti
-
m
o
d
al
f
ea
tu
r
e
f
u
s
io
n
.
T
h
is
co
m
p
r
eh
en
s
iv
e
an
al
y
s
is
in
b
o
th
th
e
tem
p
o
r
al
an
d
f
r
eq
u
e
n
cy
d
o
m
ain
s
en
h
an
ce
s
th
e
ac
cu
r
ac
y
a
n
d
r
o
b
u
s
tn
ess
o
f
ar
r
h
y
t
h
m
ia
d
etec
tio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
2
,
Au
g
u
s
t
20
25
:
1
1
6
4
-
1
175
1166
−
Ad
v
an
ce
d
d
ee
p
atten
tio
n
n
e
two
r
k
-
b
ased
f
ea
tu
r
e
ex
t
r
ac
tio
n
:
lev
er
ag
in
g
th
e
p
o
wer
o
f
d
ee
p
atten
tio
n
n
etwo
r
k
m
o
d
els,
DANI
N
e
x
ce
ls
in
r
ec
o
g
n
izin
g
an
d
e
x
tr
ac
tin
g
c
o
m
p
lex
p
atter
n
s
an
d
lo
n
g
-
r
an
g
e
d
ep
en
d
e
n
cies in
E
C
G
d
ata.
T
h
is
ad
v
an
c
ed
f
ea
tu
r
e
ex
tr
ac
tio
n
ca
p
ab
ilit
y
s
ig
n
if
ican
tly
im
p
r
o
v
es th
e
m
o
d
el
’
s
ab
ilit
y
to
ac
cu
r
ately
class
if
y
v
ar
io
u
s
ty
p
es o
f
a
r
r
h
y
th
m
ias.
−
E
n
h
an
ce
d
in
ter
p
r
eta
b
ilit
y
an
d
u
s
ab
ilit
y
:
b
y
in
c
o
r
p
o
r
atin
g
an
in
f
e
r
en
ce
m
o
d
el
s
y
s
tem
,
th
e
DANI
N
m
eth
o
d
o
l
o
g
y
n
o
t
o
n
ly
b
o
o
s
t
d
iag
n
o
s
tic
ac
cu
r
ac
y
b
u
t
also
im
p
r
o
v
es
th
e
in
ter
p
r
etab
ilit
y
o
f
t
h
e
r
esu
lts
.
T
h
i
s
en
s
u
r
es
th
at
th
e
m
o
d
els
ar
e
p
r
ac
tical
an
d
r
eliab
le
f
o
r
clin
ical
u
s
e,
f
ac
ilit
atin
g
b
etter
d
ec
is
i
o
n
-
m
ak
i
n
g
an
d
p
atien
t c
ar
e.
T
h
e
p
ap
e
r
is
o
r
g
an
ized
in
to
4
s
ec
tio
n
s
;
th
e
f
ir
s
t
s
ec
tio
n
g
i
v
es
a
b
r
ief
i
n
tr
o
d
u
ctio
n
t
o
ar
r
h
y
th
m
ia
th
e
s
ec
o
n
d
s
ec
tio
n
g
iv
es
a
th
o
r
o
u
g
h
liter
atu
r
e
s
u
r
v
ey
.
T
h
e
t
h
ir
d
s
ec
tio
n
d
ef
in
es
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
.
T
h
e
f
o
u
r
th
s
ec
tio
n
d
eter
m
in
es th
e
p
e
r
f
o
r
m
an
ce
ev
alu
atio
n
wh
er
e
t
h
e
r
e
s
u
lts
ar
e
g
iv
en
in
th
e
f
o
r
m
o
f
g
r
ap
h
s
a
n
d
tab
les.
2.
RE
L
AT
E
D
WO
RK
E
C
Gs
h
av
e
b
ee
n
class
if
ied
u
s
in
g
DNNs
in
th
e
r
ec
en
t
p
as
t.
DNNs
m
ay
d
ir
ec
tly
d
er
iv
e
a
f
ea
t
u
r
e
ex
tr
ac
tio
n
f
u
n
ctio
n
f
r
o
m
th
e
r
aw
in
p
u
t d
ata
b
y
u
tili
zin
g
th
e
d
ataset
’
s
p
r
o
b
ab
ilit
y
d
is
tr
ib
u
ti
o
n
,
th
is
is
h
o
w
th
ese
ap
p
r
o
ac
h
es
ar
e
d
if
f
er
e
n
t
f
r
o
m
tr
ad
itio
n
al
o
n
es.
Featu
r
es
d
er
iv
e
d
f
r
o
m
a
DNN
m
o
d
el
ca
n
b
e
m
o
r
e
co
m
p
r
eh
e
n
s
iv
e
th
a
n
f
ea
tu
r
es
p
r
o
d
u
ce
d
m
an
u
ally
wh
e
n
a
la
r
g
e
e
n
o
u
g
h
q
u
a
n
tity
o
f
tr
ai
n
in
g
d
ata
is
av
ailab
le.
Ven
tr
icu
lar
ar
r
h
y
th
m
ias
wer
e
id
en
tifie
d
b
y
tr
ain
in
g
an
ap
p
r
o
p
r
iate
f
ea
tu
r
e
m
a
p
p
in
g
with
a
s
tack
ed
d
en
o
is
in
g
au
to
en
co
d
er
(
SDAE
)
.
T
h
en
,
b
y
ad
d
in
g
a
So
f
tMa
x
r
eg
r
ess
io
n
lay
er
to
th
e
h
id
d
en
r
ep
r
esen
tatio
n
lay
er
,
DNNs
ar
e
u
s
ed
[
6
]
.
Au
to
m
atic
id
en
t
if
icatio
n
o
f
ca
r
d
iac
ar
r
h
y
th
m
i
as
h
as
b
ee
n
m
ad
e
p
o
s
s
ib
le
b
y
th
e
p
ar
allel
u
s
e
o
f
C
NNs,
b
y
u
s
in
g
C
NNs
to
id
en
tify
AF.
T
h
e
ap
p
licatio
n
o
f
a
m
u
ltis
ca
le
f
u
s
io
n
o
f
d
ee
p
c
o
n
v
o
lu
ti
o
n
al
n
e
u
r
al
n
etwo
r
k
s
(
MS
-
C
NN)
h
as
b
ee
n
s
u
g
g
ested
as
a
s
o
lu
tio
n
to
th
e
AF
p
r
o
b
lem
[
7
]
.
B
y
u
s
in
g
f
ilter
s
o
f
v
ar
io
u
s
s
izes,
th
e
m
eth
o
d
m
ak
es
u
s
e
o
f
a
two
-
s
tr
ea
m
co
n
v
o
lu
tio
n
a
l
n
etwo
r
k
ar
ch
itectu
r
e
to
ex
tr
ac
t
in
f
o
r
m
atio
n
at
n
u
m
er
o
u
s
s
ca
les.
Dek
im
p
e
an
d
B
o
l
[
8
]
,
a
C
NN
u
s
in
g
th
e
r
esid
u
al
n
etwo
r
k
d
esig
n
was
c
r
ea
ted
to
p
r
ec
is
ely
ca
teg
o
r
ize
1
2
r
h
y
th
m
class
es.
I
n
th
e
f
ield
o
f
ar
r
h
y
th
m
ia
class
if
icatio
n
,
C
NN
s
ar
e
f
r
e
q
u
en
tly
u
tili
ze
d
to
class
if
y
ar
r
h
y
th
m
ias
at
th
e
b
e
at
lev
el.
I
n
th
ese
k
i
n
d
s
o
f
s
it
u
atio
n
s
,
th
e
m
o
d
el
’
s
in
p
u
t
d
a
ta
is
u
s
u
ally
m
u
ch
s
h
o
r
ter
,
f
r
e
q
u
en
tly
n
u
m
b
er
in
g
ju
s
t
in
th
e
h
u
n
d
r
ed
s
o
f
s
am
p
l
es
[
9
]
.
A
n
in
e
-
la
y
er
C
NN
ex
am
p
le
was
cr
ea
ted
to
au
to
m
atica
lly
r
ec
o
g
n
ize
an
d
c
ateg
o
r
ize
f
iv
e
d
is
t
in
ct
k
in
d
s
o
f
h
ea
r
t
b
ea
ts
.
T
wo
m
o
r
e
n
etw
o
r
k
to
p
o
l
o
g
ies
th
at
ar
e
o
f
ten
u
s
ed
in
th
e
f
ield
o
f
E
C
G
class
if
icatio
n
ar
e
th
e
r
estricte
d
B
o
ltzm
an
n
m
ac
h
i
n
e
(
R
B
M)
an
d
th
e
au
to
en
co
d
er
.
A
u
n
i
q
u
e
ap
p
r
o
a
ch
b
ased
o
n
DL
is
u
s
ed
as
a
s
o
lu
tio
n
to
th
e
p
r
ev
io
u
s
ly
d
escr
ib
ed
p
r
o
b
lem
.
T
h
e
p
r
ev
io
u
s
m
et
h
o
d
i
n
teg
r
ates
a
SVM
with
an
au
to
en
c
o
d
er
n
etwo
r
k
th
at
u
s
es
L
STM
ar
ch
itectu
r
e
to
class
if
y
ar
r
h
y
th
m
ias
[
1
0
]
.
Acc
o
r
d
in
g
to
r
esear
ch
,
C
NN
an
d
L
ST
M
m
ay
b
e
in
teg
r
ated
to
au
t
o
m
atica
lly
class
if
y
ar
r
h
y
th
m
ias.
T
h
e
class
if
icati
o
n
p
er
f
o
r
m
a
n
ce
u
ti
lizin
g
v
a
r
io
u
s
r
ec
o
r
d
in
g
tim
es
f
o
r
E
C
G
d
ata
was
al
s
o
in
v
esti
g
ated
in
th
is
wo
r
k
.
Usi
n
g
an
en
s
em
b
le
n
etwo
r
k
m
o
d
el
b
ased
o
n
DL
im
p
r
o
v
ed
th
e
p
er
f
o
r
m
an
ce
o
f
a
s
in
g
le
n
etwo
r
k
.
T
h
r
ee
d
if
f
er
e
n
t
n
etwo
r
k
s
ar
e
in
co
r
p
o
r
ated
in
to
th
e
m
o
d
el
’
s
ar
ch
itectu
r
e
to
r
ec
o
g
n
ize
a
n
d
g
ath
er
d
ata.
T
h
e
p
r
ev
io
u
s
ly
o
u
tlin
ed
p
r
o
ce
d
u
r
e
y
ield
s
an
e
x
tr
em
ely
ef
f
icie
n
t
m
eth
o
d
f
o
r
d
ata
id
en
tific
atio
n
an
d
g
ath
e
r
in
g
.
Acc
o
r
d
in
g
to
p
r
elim
in
ar
y
r
ese
ar
ch
,
s
ev
er
al
alg
o
r
ith
m
s
h
av
e
d
em
o
n
s
tr
ated
p
o
ten
tial
i
n
th
e
au
to
m
ated
ca
teg
o
r
izatio
n
o
f
ar
r
h
y
t
h
m
ias
u
s
in
g
E
C
G
d
ata.
B
ef
o
r
e
th
ese
alg
o
r
ith
m
s
ar
e
s
u
cc
ess
f
u
lly
a
p
p
lied
in
r
ea
l
-
wo
r
ld
cir
cu
m
s
tan
ce
s
,
s
ev
er
al
ch
allen
g
es
n
ee
d
to
b
e
r
eso
lv
ed
.
I
t
is
n
o
tewo
r
th
y
th
at
th
e
f
ea
tu
r
es
o
f
o
b
tain
e
d
in
d
iv
id
u
al
E
C
Gs m
ig
h
t y
ield
v
alu
ab
le
clin
ical
d
ata
f
o
r
au
to
m
ated
ca
r
d
iac
ar
r
h
y
t
h
m
ia
id
en
tific
atio
n
.
Ho
wev
er
,
it
’
s
cr
u
cial
to
r
em
em
b
er
t
h
at
E
C
G
s
ig
n
als
f
r
o
m
p
eo
p
le
with
v
ar
io
u
s
m
ed
ical
d
is
o
r
d
er
s
f
r
eq
u
en
tly
h
av
e
u
n
iq
u
e
tem
p
o
r
al
an
d
m
o
r
p
h
o
lo
g
ical
f
ea
tu
r
es
[
1
1
]
.
I
n
d
iv
id
u
al
d
if
f
er
en
ce
s
m
ig
h
t
ca
u
s
e
ea
ch
p
er
s
o
n
’
s
E
C
G
s
ig
n
al
to
f
lu
ctu
ate
d
if
f
e
r
en
tly
o
v
er
ti
m
e.
Fu
r
th
er
m
o
r
e,
e
v
en
p
eo
p
le
with
th
e
s
am
e
m
ed
ical
c
o
n
d
itio
n
c
o
u
ld
h
av
e
d
if
f
er
en
t
E
C
G
m
o
r
p
h
o
lo
g
ies.
I
t
is
ess
en
tial
to
r
em
em
b
er
th
at
d
if
f
er
e
n
t
h
ea
r
t
d
is
ea
s
es
m
ig
h
t
p
r
esen
t
with
id
en
tical
E
C
G
f
ea
tu
r
es.
On
e
m
ajo
r
o
b
s
tacle
is
an
aly
zin
g
an
d
e
x
tr
ac
tin
g
c
h
ar
ac
ter
is
tics
to
d
etec
t
ca
r
d
iac
d
is
ea
s
es.
T
h
e
d
is
tin
ct
ch
ar
ac
ter
is
tics
o
f
ea
ch
p
atien
t
’
s
r
h
y
t
h
m
th
at
m
ay
v
ar
y
f
r
o
m
t
h
e
t
r
ain
in
g
s
et
ar
e
n
o
t
tak
en
in
to
ac
co
u
n
t
b
y
th
e
ar
r
h
y
th
m
ia
class
if
icatio
n
alg
o
r
ith
m
s
cu
r
r
en
tly
in
u
s
e.
T
h
ese
alg
o
r
ith
m
s
m
ak
e
u
s
e
o
f
r
elativ
ely
tin
y
tr
ain
in
g
d
atasets
.
As
a
r
esu
lt,
it
’
s
p
o
s
s
ib
le
th
at
th
e
ex
is
tin
g
tech
n
iq
u
es
wo
n
’
t
wo
r
k
as
well
i
n
p
r
ac
tical
s
itu
atio
n
s
.
T
h
e
g
ath
er
in
g
o
f
a
p
atien
t
’
s
lo
n
g
-
ter
m
E
C
G
r
ec
o
r
d
s
is
m
ad
e
ea
s
ier
b
y
lo
n
g
-
ter
m
E
C
G
m
o
n
ito
r
in
g
eq
u
ip
m
en
t,
wh
ich
h
el
p
s
to
ad
d
r
ess
th
is
p
r
o
b
le
m
.
W
ith
th
ese
g
ad
g
ets,
au
to
m
ated
ca
teg
o
r
izatio
n
m
eth
o
d
s
ar
e
a
p
p
lied
[
1
2
]
.
Kir
an
y
az
’
s
r
ea
l
-
tim
e
p
atien
t
-
s
p
ec
if
ic
E
C
G
class
if
icatio
n
tec
h
n
iq
u
e
is
th
e
o
n
l
y
way
to
cla
s
s
if
y
lo
n
g
-
ter
m
E
C
G
r
ec
o
r
d
in
g
s
o
f
a
p
at
ien
t;
Z
h
an
g
p
r
o
p
o
s
ed
a
p
atien
t
-
s
p
ec
if
ic
E
C
G
cla
s
s
if
icatio
n
m
eth
o
d
u
s
in
g
R
NN
to
class
if
y
E
C
G
b
ea
t
s
with
d
i
f
f
er
en
t
h
ea
r
t
r
ates
an
d
ca
p
tu
r
e
tem
p
o
r
al
co
r
r
elatio
n
f
r
o
m
E
C
G
s
ig
n
al
s
am
p
les.
1
D
C
NNs
s
er
v
e
as
th
e
f
o
u
n
d
atio
n
f
o
r
th
is
m
eth
o
d
.
A
te
ch
n
iq
u
e
f
o
r
ca
teg
o
r
izin
g
f
i
v
e
t
y
p
ical
k
in
d
s
o
f
ar
r
h
y
th
m
ia
s
ig
n
als
u
s
in
g
a
o
n
e
-
d
im
en
s
io
n
al
C
NN
(
1
D
-
C
N
N)
[
1
3
]
.
A
d
ee
p
two
-
d
im
en
s
i
o
n
al
C
NN
was
u
s
ed
b
y
J
u
n
as
a
u
s
ef
u
l
m
eth
o
d
f
o
r
id
e
n
tify
in
g
E
C
G
ar
r
h
y
t
h
m
ias.
I
n
th
e
ar
ea
o
f
p
atter
n
r
ec
o
g
n
itio
n
,
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
d
va
n
ce
d
d
ee
p
a
tten
tio
n
n
eu
r
a
l in
feren
ce
n
etw
o
r
k
fo
r
en
h
a
n
ce
d
a
r
r
h
yth
mia
…
(
H.
S
u
mit
h
a
)
1167
af
o
r
em
en
tio
n
ed
n
eu
r
al
n
etwo
r
k
h
as
d
e
m
o
n
s
tr
a
ted
e
x
ce
p
ti
o
n
al
p
er
f
o
r
m
a
n
ce
.
T
h
r
o
u
g
h
t
h
e
u
s
e
o
f
tr
an
s
f
er
lear
n
in
g
f
r
o
m
2
D
d
ee
p
C
NN
f
ea
tu
r
es,
th
is
m
eth
o
d
was
a
b
le
to
class
if
y
E
C
G
d
is
o
r
d
er
s
.
T
h
e
m
eth
o
d
o
lo
g
y
o
u
tlin
ed
was
em
p
l
o
y
ed
to
id
e
n
tify
an
d
class
if
y
f
o
u
r
d
is
cr
ete
E
C
G
p
atter
n
s
.
I
n
th
eir
s
tu
d
y
,
[
1
4
]
i
n
t
r
o
d
u
ce
d
a
n
o
v
el
tech
n
i
q
u
e
f
o
r
ca
r
r
y
in
g
o
u
t
f
ea
tu
r
e
e
x
tr
ac
tio
n
ac
r
o
s
s
m
an
y
d
o
m
ain
s
.
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
co
m
b
in
es
W
T
with
k
er
n
el
-
I
C
A.
T
h
e
d
at
a
was
co
m
p
r
ess
ed
u
s
in
g
PC
A.
T
h
e
I
C
A
ap
p
r
o
ac
h
was
ap
p
li
ed
to
th
e
E
C
G
d
ata
to
d
etec
t
f
ea
tu
r
es
an
d
m
in
im
ize
d
im
en
s
io
n
a
lity
[
1
5
]
.
Sele
ctin
g
s
p
ec
if
ic
tr
aits
i
s
s
o
m
eti
m
es
b
ased
o
n
th
e
ev
alu
atio
n
o
f
ex
p
er
ien
ce
.
Am
o
n
g
th
e
ch
ar
ac
ter
is
tics
ar
e
h
ig
h
er
-
o
r
d
er
s
p
ec
tr
a,
lin
ea
r
an
d
n
o
n
lin
ea
r
f
ea
tu
r
es,
s
p
ar
s
e
f
ea
tu
r
es,
en
tr
o
p
y
-
b
ased
f
ea
tu
r
es,
an
d
s
tatis
tical
f
ea
tu
r
es.
3.
P
RO
P
O
SE
D
M
E
T
H
O
D
T
h
e
p
r
o
p
o
s
ed
DANI
N
m
eth
o
d
o
lo
g
y
c
o
n
s
is
ts
o
f
a
d
etailed
ac
co
u
n
t
o
f
th
e
p
r
o
ce
d
u
r
es
an
d
tech
n
iq
u
es
em
p
lo
y
ed
t
o
co
llect
d
ata
an
d
tak
e
E
C
Gs.
I
n
itially
,
th
e
p
r
o
ce
s
s
o
f
co
n
v
er
tin
g
E
C
G
im
p
u
ls
es
in
to
s
p
ec
tr
o
g
r
am
s
is
an
aly
ze
d
,
an
d
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
in
te
g
r
ates
o
n
e
-
d
im
en
s
io
n
al
E
C
G
d
ata
with
th
e
c
o
r
r
esp
o
n
d
in
g
two
-
d
im
en
s
io
n
al
s
p
ec
tr
al
im
ag
es
u
s
in
g
m
u
ltimo
d
al
f
ea
tu
r
e
f
u
s
io
n
.
B
ec
au
s
e
o
f
th
is
in
teg
r
atio
n
,
th
e
d
ata
is
co
llected
in
b
o
th
th
e
tem
p
o
r
al
an
d
f
r
eq
u
en
cy
d
o
m
ai
n
s
,
wh
i
ch
allo
ws
a
co
m
p
r
eh
e
n
s
iv
e
u
n
d
er
s
tan
d
in
g
o
f
th
e
ca
r
d
iac
s
ig
n
als.
Ad
d
itio
n
ally
,
a
d
ee
p
atten
tio
n
n
etwo
r
k
-
b
a
s
ed
f
ea
tu
r
e
ex
tr
ac
tio
n
en
c
o
d
e
r
was
s
p
ec
if
ically
d
ev
elo
p
e
d
to
e
x
tr
ac
t
f
ea
tu
r
es
f
r
o
m
o
n
e
-
d
im
en
s
io
n
al
E
C
G
d
ata
s
ig
n
als
an
d
s
p
ec
tr
u
m
im
a
g
es.
Dee
p
atten
tio
n
n
etwo
r
k
s
ar
e
o
f
ten
em
p
lo
y
ed
in
s
ev
er
al
d
o
m
ain
s
,
in
clu
d
in
g
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
.
Ho
wev
er
,
th
eir
ap
p
licatio
n
in
E
C
G
s
ig
n
al
an
al
y
s
is
is
n
o
v
el,
p
ar
ticu
lar
ly
wh
e
n
in
teg
r
ated
with
i
n
f
er
en
c
e
m
o
d
el
lo
g
ic.
Dee
p
atten
tio
n
n
etwo
r
k
s
ar
e
an
ec
o
n
o
m
ical
m
ea
n
s
o
f
r
e
p
r
esen
tin
g
co
m
p
lex
p
atter
n
s
a
n
d
l
o
n
g
-
r
a
n
g
e
co
n
n
ec
tio
n
s
in
E
C
G
d
ata.
Fu
r
t
h
er
m
o
r
e
,
a
m
o
r
e
ac
c
u
r
ate
ca
te
g
o
r
izatio
n
a
n
d
f
o
r
ec
asti
n
g
o
f
E
C
G
d
ata
is
ca
r
r
ied
o
u
t
b
y
ex
am
in
in
g
th
e
co
m
b
in
ed
m
u
lti
-
m
o
d
al
attr
ib
u
tes
th
r
o
u
g
h
th
e
u
tili
za
tio
n
o
f
an
in
f
er
en
ce
m
o
d
el.
T
h
e
in
teg
r
atio
n
o
f
th
e
in
f
er
en
ce
m
o
d
el
s
y
s
tem
o
v
er
c
o
m
es
a
m
ajo
r
b
ar
r
ier
in
ap
p
ly
in
g
DL
to
m
ed
ical
d
ata
an
d
en
h
an
ce
s
th
e
ac
cu
r
ac
y
o
f
th
e
m
o
d
el
’
s
p
r
e
d
ictio
n
s
.
T
h
e
in
f
er
en
ce
m
o
d
el
m
o
d
u
le
en
h
a
n
ce
s
th
e
d
iag
n
o
s
tic
p
r
o
ce
d
u
r
e
’
s
r
o
b
u
s
tn
ess
b
y
p
r
o
v
id
in
g
a
m
ea
n
s
to
h
an
d
le
th
e
in
h
er
e
n
t
u
n
ce
r
tain
ties
a
n
d
v
ar
iab
ilit
ies
in
E
C
G
r
ea
d
in
g
s
.
Fig
u
r
e
1
p
r
esen
ts
th
e
co
m
p
r
e
h
en
s
iv
e
s
ch
em
atic
o
f
th
e
m
o
d
el
’
s
f
r
am
e.
Fig
u
r
e
1
.
Pro
p
o
s
ed
DANI
N
a
r
ch
itectu
r
e
3
.
1
.
E
f
f
icient
s
ig
na
l t
ra
ns
f
o
r
m
a
t
io
n
T
h
e
f
u
n
d
a
m
en
tal
m
o
d
u
le
o
f
th
e
win
d
o
wed
Fo
u
r
ier
tr
an
s
f
o
r
m
(
W
FT)
is
to
d
ec
o
m
p
o
s
e
th
e
s
ig
n
al
in
to
an
u
n
lim
ited
n
u
m
b
er
o
f
s
m
aller
s
eg
m
en
ts
an
d
s
u
b
s
eq
u
en
tl
y
p
er
f
o
r
m
win
d
o
w
p
r
o
ce
s
s
in
g
to
ea
ch
o
f
th
ese
s
eg
m
en
ts
.
T
h
e
win
d
o
wed
s
m
a
ll
s
eg
m
en
t
s
ig
n
al
m
ay
b
e
a
n
al
y
ze
d
u
s
in
g
th
e
Fo
u
r
ier
tr
a
n
s
f
o
r
m
s
u
b
s
eq
u
e
n
tly
it
is
co
n
s
id
er
ed
as a
s
tatio
n
ar
y
s
ig
n
al.
Ass
u
m
e
th
e
s
ig
n
al
is
(
)
,
th
e
tr
an
s
f
o
r
m
atio
n
o
f
th
e
s
ig
n
al
is
th
en
ap
p
lied
as g
iv
en
in
(
1
)
.
(
,
ℎ
)
=
∫
(
)
(
−
)
−
2
ℎ
(
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
2
,
Au
g
u
s
t
20
25
:
1
1
6
4
-
1
175
1168
W
h
er
e
in
ℎ
is
f
r
eq
u
en
cy
an
d
(
)
is
th
e
win
d
o
w
f
u
n
ctio
n
,
f
o
r
f
u
r
t
h
er
p
r
o
ce
s
s
in
g
o
f
th
e
Fo
u
r
ier
t
r
an
s
f
o
r
m
o
n
th
e
s
ig
n
al
f
o
r
th
e
lo
ca
l
tim
e
-
f
r
eq
u
e
n
cy
p
la
n
e,
a
lim
ited
win
d
o
w
f
u
n
ctio
n
(
)
is
ad
d
ed
b
ef
o
r
e
th
e
Fo
u
r
ier
tr
an
s
f
o
r
m
s
s
o
th
is
f
u
n
ctio
n
s
eg
m
en
ts
th
e
n
o
n
-
s
tatio
n
ar
y
s
ig
n
al
in
to
u
n
li
m
ited
n
o
o
f
s
eg
m
en
ts
,
with
in
th
e
ac
tio
n
o
f
th
e
win
d
o
w
f
u
n
ctio
n
(
)
,
th
e
Fo
u
r
ier
tr
an
s
f
o
r
m
atio
n
is
p
er
f
o
r
m
ed
o
n
ea
c
h
s
eg
m
e
n
t
o
f
th
e
s
ig
n
al
h
en
ce
o
b
tain
in
g
th
e
lo
ca
l
s
p
ec
tr
u
m
d
if
f
er
en
c
e
at
v
ar
io
u
s
tim
es.
T
h
e
f
r
eq
u
en
cy
ass
o
ciate
d
with
ea
ch
s
eg
m
en
t
o
f
th
e
s
ig
n
al
d
eter
m
in
es
th
e
a
ctio
n
o
f
th
e
win
d
o
w
f
u
n
ctio
n
(
)
,
th
e
s
m
aller
it
is
h
ig
h
er
its
r
e
s
o
lu
tio
n
is
,
th
e
tim
e
an
d
b
an
d
wid
t
h
o
f
th
e
w
in
d
o
w
ar
e
co
n
tr
ad
icto
r
y
an
d
th
e
lim
it
s
ize
o
f
th
e
win
d
o
w
is
ar
b
itra
r
ily
v
er
y
s
m
all,
o
n
ce
af
ter
win
d
o
win
g
t
h
e
s
ig
n
al
(
)
th
e
Fo
u
r
ier
tr
an
s
f
o
r
m
is
f
u
r
th
er
co
n
n
ec
ted
to
tim
e
v
ar
iab
le.
T
h
e
win
d
o
w
f
u
n
ctio
n
is
ev
alu
ated
b
y
th
e
(
2
)
.
(
,
)
(
)
=
(
−
)
,
(
)
∈
2
(
)
(
2
)
Her
e
d
en
o
tes
th
e
f
r
eq
u
en
cy
s
h
if
t
p
ar
am
eter
f
o
r
th
e
win
d
o
w
wh
er
ea
s
d
en
o
tes
th
e
tim
e
s
h
if
t
p
ar
am
et
e
r
win
d
o
w,
b
y
ass
u
m
in
g
o
p
er
ati
o
n
s
lik
e
=
an
d
=
,
th
is
s
ig
n
al
tr
an
s
f
o
r
m
atio
n
is
r
ep
r
esen
ted
b
y
(
3
)
.
W
h
er
e
(
)
is
th
e
co
ef
f
icien
t
o
f
th
e
W
FT
tr
an
s
f
o
r
m
,
f
r
o
m
th
e
ab
o
v
e
eq
u
atio
n
wh
er
ein
,
(
,
B
)
tr
an
s
f
o
r
m
s
th
e
s
ig
n
al
as
a
s
eg
m
en
ted
s
er
ies
o
f
r
ec
tan
g
u
lar
e
q
u
al
b
lo
ck
s
at
eq
u
al
in
ter
v
als
with
in
th
e
v
ar
iab
le
,
.
T
h
e
eq
u
al
r
eso
lu
tio
n
o
f
th
e
s
ig
n
al
m
ap
p
ed
b
y
2
(
)
-
2
(
2
)
f
o
r
th
e
ti
m
e
-
f
r
eq
u
en
cy
p
lan
e
is
an
ess
en
tial f
ea
tu
r
e
o
f
th
e
W
FT
tr
an
s
f
o
r
m
.
,
(
)
=
{
,
,
}
=
{
,
,
}
(
3
)
3
.
2
.
Dee
p
a
t
t
ent
io
n net
wo
rk
-
ba
s
ed
f
ea
t
ure
ex
t
ra
ct
i
o
n
I
n
th
is
s
ec
tio
n
,
th
e
an
aly
s
is
o
f
th
e
s
tr
u
ctu
r
e
o
f
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
m
o
d
u
le
is
ca
r
r
ied
o
u
t.
Dee
p
atten
tio
n
n
etwo
r
k
d
esig
n
s
ar
e
th
e
b
asis
o
f
th
is
tech
n
iq
u
e
f
o
r
r
ec
o
g
n
izin
g
an
d
ex
tr
ac
ti
n
g
cr
u
cial
in
f
o
r
m
atio
n
f
r
o
m
s
p
ec
tr
al
im
ag
es
an
d
E
C
G
s
ig
n
als.
T
h
is
b
im
o
d
al
ap
p
r
o
ac
h
u
tili
ze
s
th
e
d
ee
p
atten
tio
n
n
etwo
r
k
’
s
a
b
ilit
y
to
d
is
tin
g
u
is
h
co
m
p
lex
p
atter
n
s
an
d
d
is
tan
t
co
n
n
ec
tio
n
s
f
r
o
m
th
e
in
p
u
t,
in
ter
p
r
etin
g
it
as
p
a
r
ticu
lar
ly
v
alu
ab
le
f
o
r
d
ec
o
d
in
g
th
e
co
m
p
lex
s
tr
u
ctu
r
es
o
b
s
er
v
ed
in
s
p
ec
tr
al
im
ag
es
an
d
E
C
G
d
ata.
T
h
e
d
ee
p
atten
tio
n
n
etw
o
r
k
m
o
d
el
h
as
ac
h
iev
e
d
s
ig
n
if
ic
an
t
u
s
e
d
u
e
t
o
its
ef
f
ec
tiv
e
n
ess
in
h
an
d
lin
g
s
eq
u
e
n
tial
d
ata
s
in
ce
its
in
itial
d
ev
elo
p
m
e
n
t
f
o
r
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
ap
p
licatio
n
s
.
Dee
p
atten
tio
n
n
etwo
r
k
-
b
ased
m
o
d
els
s
u
r
p
ass
co
m
p
u
ter
v
is
io
n
task
s
s
u
ch
as
class
if
icatio
n
,
d
etec
tio
n
,
an
d
s
eg
m
en
tatio
n
,
as
well
as
ad
d
r
ess
in
g
ch
allen
g
es
in
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
.
Fig
u
r
e
1
illu
s
tr
ates
th
e
s
tr
u
ctu
r
e
o
f
t
h
e
ty
p
ical
d
ee
p
atten
ti
o
n
n
etwo
r
k
,
wh
ich
co
n
s
is
ts
o
f
a
g
r
o
u
p
o
f
en
co
d
er
s
an
d
a
g
r
o
u
p
o
f
d
ec
o
d
er
s
.
T
h
e
s
elf
-
atten
tio
n
m
ec
h
a
n
is
m
is
th
e
m
o
s
t
e
s
s
en
tial
co
m
p
o
n
e
n
t
in
b
o
th
th
e
en
co
d
er
an
d
d
ec
o
d
e
r
.
T
h
e
d
e
ep
atten
tio
n
n
etwo
r
k
m
o
d
el
u
tili
ze
s
th
e
s
elf
-
atten
tio
n
ap
p
r
o
ac
h
to
ac
cu
r
atel
y
d
etec
t
g
lo
b
al
r
elatio
n
s
h
ip
s
ac
r
o
s
s
d
if
f
er
en
t
r
eg
i
o
n
s
.
Dee
p
atten
tio
n
n
etwo
r
k
s
u
r
p
a
s
s
es
C
NN
an
d
R
NN
in
ad
d
r
ess
in
g
s
eq
u
en
ce
p
r
o
b
lem
s
d
u
e
to
th
e
s
am
e
u
n
d
e
r
ly
in
g
c
au
s
e.
T
h
e
s
elf
-
atten
tio
n
m
ec
h
an
is
m
,
wh
ich
is
th
e
m
ain
m
ec
h
a
n
is
m
o
f
th
e
m
o
d
el,
allo
ws
it
to
p
r
io
r
itize
d
if
f
e
r
e
n
t
p
ar
ts
o
f
t
h
e
in
p
u
t
d
ata.
T
h
is
en
ab
les
th
e
m
o
d
el
to
r
ed
u
ce
th
e
im
p
ac
t
o
f
less
s
ig
n
if
ican
t
ch
ar
ac
ter
is
tics
an
d
co
n
ce
n
tr
ate
o
n
th
e
s
ig
n
if
ica
n
t
o
n
es.
T
h
e
d
ee
p
atten
tio
n
n
e
two
r
k
ar
ch
itectu
r
e
is
ad
ap
ted
f
o
r
s
u
itab
le
an
aly
s
is
,
en
ab
lin
g
u
s
to
an
aly
ze
b
o
th
two
-
d
im
en
s
io
n
a
l
s
p
ec
tr
al
im
ag
es
an
d
o
n
e
-
d
i
m
en
s
io
n
al
E
C
G
d
ata.
T
h
e
m
o
d
el
co
n
s
is
ts
o
f
two
p
ar
a
llel
b
r
an
ch
es,
ea
ch
s
p
ec
ialized
f
o
r
p
r
o
ce
s
s
in
g
a
d
is
tin
ct
d
ata
m
o
d
ality
.
T
h
e
n
ex
t,
s
tep
in
th
is
p
r
o
ce
s
s
is
ca
lcu
l
atin
g
d
ata
f
o
r
b
o
th
E
C
G
s
ig
n
als
an
d
th
e
co
r
r
esp
o
n
d
in
g
s
p
ec
tr
al
im
ag
es
in
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
m
o
d
el
b
ased
o
n
d
ee
p
atten
tio
n
n
etw
o
r
k
s
.
T
h
is
p
r
o
ce
s
s
co
n
cl
u
d
es
with
th
e
in
teg
r
atio
n
o
f
c
h
ar
ac
ter
is
tics
.
T
h
e
o
b
jectiv
e
o
f
th
is
p
r
o
ce
s
s
is
to
ex
tr
ac
t
an
d
m
er
g
e
th
e
m
o
s
t
ess
en
tial
d
ata
f
r
o
m
b
o
th
m
o
d
alities
f
o
r
f
u
r
th
er
a
n
aly
s
is
u
s
in
g
a
s
er
ies
o
f
co
m
p
u
ter
p
h
ases
.
Fo
r
th
e
E
C
G
s
ig
n
al
an
al
y
s
is
,
th
e
d
ee
p
atten
tio
n
n
etwo
r
k
p
r
o
ce
s
s
es
th
e
in
p
u
t
s
ig
n
al
co
n
s
is
tin
g
o
f
d
ata
p
o
in
ts
r
esu
ltin
g
in
a
s
eq
u
en
ce
o
f
f
ea
tu
r
e
v
ec
t
o
r
s
=
[
1
,
2
,
…
,
]
.
E
ac
h
is
a
f
ea
tu
r
e
d
en
o
ted
b
y
a
tim
e
s
tep
,
th
e
co
m
p
u
tatio
n
is
d
en
o
ted
at
th
e
d
ee
p
atten
tio
n
n
etwo
r
k
lay
er
f
o
r
th
e
E
C
G
b
r
an
ch
is
d
en
o
ted
b
y
th
e
s
elf
-
atten
tio
n
m
ec
h
a
n
is
m
d
ef
in
ed
as
s
h
o
wn
in
(
4
)
.
T
h
e
s
elf
-
atten
tio
n
m
ec
h
a
n
is
m
ev
al
u
ates
th
e
atten
tio
n
s
co
r
es a
t v
ar
io
u
s
p
o
s
itio
n
s
o
f
th
e
in
p
u
t seq
u
en
ce
to
f
o
cu
s
o
n
th
e
in
ter
d
ep
en
d
en
cies ir
r
esp
ec
tiv
e
o
f
th
e
p
o
s
iti
on
as
f
o
llo
ws
in
(
4
)
.
Her
e
,
is
th
e
q
u
er
y
,
k
ey
,
an
d
v
alu
e
m
atr
ice
s
d
er
iv
ed
th
r
o
u
g
h
th
e
in
p
u
t,
a
n
d
is
th
e
k
ey
d
im
en
s
io
n
ev
al
u
ated
a
s
g
iv
en
in
(
5
)
.
(
,
,
)
=
(
√
)
(
4
)
=
∗
=
∗
(
5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
d
va
n
ce
d
d
ee
p
a
tten
tio
n
n
eu
r
a
l in
feren
ce
n
etw
o
r
k
fo
r
en
h
a
n
ce
d
a
r
r
h
yth
mia
…
(
H.
S
u
mit
h
a
)
1169
=
∗
,
,
ar
e
weig
h
t
m
atr
ices
r
an
d
o
m
l
y
in
itialized
at
th
e
in
itial
o
f
th
e
n
etwo
r
k
tr
ain
in
g
,
th
eir
v
al
u
e
is
ad
ju
s
ted
as
th
e
b
ac
k
p
r
o
p
a
g
atio
n
n
etwo
r
k
a
n
d
r
is
g
iv
en
as
th
e
in
p
u
t
to
th
e
s
elf
-
atten
tio
n
lay
er
.
E
n
h
a
n
ce
d
atten
tio
n
m
ec
h
a
n
is
m
(
E
AM
)
is
th
e
ex
ten
s
io
n
o
f
th
e
atten
tio
n
m
ec
h
a
n
is
m
th
at
u
s
es
m
u
ltip
le
p
ar
allel
atten
tio
n
to
en
h
an
ce
th
e
m
o
d
el
’
s
ab
ilit
y
th
at
ex
tr
ac
t th
e
d
ata
f
ea
tu
r
es.
E
ac
h
h
ea
d
ca
p
tu
r
es th
e
v
ar
ied
atten
tio
n
f
o
cu
s
at
a
s
im
ilar
tim
e
asp
ec
t
th
e
E
AM
h
as
ce
r
tain
s
im
ilar
itie
s
th
at
c
ap
tu
r
e
v
ar
io
u
s
f
ea
t
u
r
es
ir
r
esp
ec
tiv
e
o
f
way
.
T
h
e
s
p
ec
if
ic
im
p
lem
en
tatio
n
h
e
r
e
i
s
a
s
im
ilar
m
ec
h
an
is
m
f
o
r
s
ev
er
al
s
ets
o
f
weig
h
t
m
atr
ices.
T
h
e
ev
al
u
atio
n
is
as
s
h
o
wn
in
(
6
)
.
Af
ter
th
is
,
th
e
o
u
tp
u
t
f
o
r
ea
ch
p
o
s
itio
n
g
o
es
th
r
o
u
g
h
a
f
ee
d
-
f
o
r
war
d
n
etw
o
r
k
ap
p
lied
at
th
e
in
d
iv
id
u
al
p
o
s
itio
n
,
th
r
o
u
g
h
th
e
p
o
s
itio
n
-
wis
e
f
ee
d
-
f
o
r
war
d
n
etwo
r
k
as g
iv
e
n
in
(
7
)
.
ℎ
=
(
,
,
)
(
,
,
)
=
(
ℎ
1
,
…
…
.
.
,
ℎ
)
(
6
)
(
)
=
(
0
,
1
+
1
)
2
+
2
(
7
)
T
h
is
ap
p
lies
to
ea
ch
elem
e
n
t
o
f
th
e
s
eq
u
en
ce
th
at
e
n
s
u
r
es
t
h
at
th
e
m
o
d
el
ca
p
tu
r
es
th
e
lo
ca
l
co
n
tex
t
with
in
ea
ch
p
o
i
n
t
o
f
t
h
e
s
ig
n
al.
Ho
wev
er
,
f
o
r
t
h
e
s
p
ec
tr
al
im
ag
e
an
aly
s
is
th
e
in
p
u
t
s
p
ec
tr
al
im
ag
e
K
is
d
en
o
ted
b
y
th
e
s
eq
u
en
ce
o
f
v
e
cto
r
s
as
=
[
1
,
2
,
…
,
]
wh
er
ein
ea
ch
co
r
r
ela
tes
to
a
p
r
o
ce
s
s
ed
r
eg
io
n
o
f
th
e
im
ag
e.
T
h
e
v
ec
to
r
s
ar
e
p
r
o
ce
s
s
ed
b
y
th
e
d
ee
p
atten
tio
n
n
etwo
r
k
s
r
elev
an
t
to
E
C
G
s
ig
n
al
an
aly
s
is
,
th
r
o
u
g
h
s
elf
-
atten
tio
n
an
d
f
ee
d
-
f
o
r
wa
r
d
n
etwo
r
k
s
as
s
p
atial
an
d
f
r
eq
u
en
c
y
-
d
o
m
ain
f
ea
tu
r
es.
Up
o
n
r
ec
eiv
i
n
g
th
e
f
ea
tu
r
e
r
e
p
r
esen
tatio
n
an
d
,
th
e
n
ex
t
p
h
ase
is
to
in
te
g
r
a
te
th
ese
f
ea
tu
r
es
with
in
an
in
teg
r
ated
r
ep
r
esen
tatio
n
th
at
ca
p
tu
r
es
th
e
i
n
s
ig
h
ts
th
r
o
u
g
h
b
o
th
tim
e
-
d
o
m
ain
an
d
f
r
eq
u
e
n
cy
-
d
o
m
ain
d
ata.
T
h
is
in
teg
r
atio
n
is
s
h
o
wn
b
elo
w
in
(
8
)
.
Alg
o
r
ith
m
1
s
h
o
ws th
e
D
ANI
N
a
lg
o
r
ith
m
.
ℎ
=
(
+
1
)
(
8
)
Alg
o
r
ith
m
1
.
Pro
p
o
s
ed
DANI
N
alg
o
r
ith
m
Input
Input:
=
{
(
,
)
}
=
1
, here
is the
−
ℎ
ECG signal along with the
relevant label;
Step 1
For each signal
in the dataset
do
Step 2
Normalize ECG signals
, through the min
-
max normalization
Step 3
Use WFT to get spectral images;
Step 4
End for
Step 5
For each normalized
,
in the dataset
do
Step 6
Use the Deep Attention Networ
k
-
based model for feature extraction through
,
Step 7
Use the Deep Attention network
-
based model for feature extraction from spectral
images;
Step 8
End for
Step 9
Features extracted are
,
1
;
Step 10
For the set of features extrac
ted the
1
do
Step 11
Integrated features through integration accordingly as to eq (11)
Step 12
Get integrated features as
ℎ
Step 13
Use adaptive input encoding (AIE) to get
ℎ
according to eq 13
Step
14
Use output extraction (OE) to get
according to eq 13
Step 15
End for
Step 16
For each step output
do
Step 17
Predict the class of the ECG signal
Step 18
End for
Step 19
Return predicted variables as {
}
=
1
output
Pre
dicted variables
for each ECG signal
4.
P
E
RF
O
RM
A
NCE
E
VA
L
U
AT
I
O
N
T
h
e
p
er
f
o
r
m
an
ce
e
v
alu
atio
n
i
s
ca
r
r
ied
o
u
t
with
th
e
ex
is
tin
g
s
tate
-
of
-
ar
t
tech
n
iq
u
es
an
d
th
e
p
r
o
p
o
s
ed
m
o
d
el
u
s
in
g
m
et
r
ics
s
u
ch
as
a
cc
u
r
ac
y
(
AC
C
)
,
p
r
ec
is
io
n
(
PR
E
)
,
r
ec
all
(
R
E
)
,
an
d
F1
-
s
co
r
e
.
T
h
e
b
aselin
e
m
eth
o
d
s
f
o
r
co
m
p
a
r
is
o
n
in
clu
d
e
r
an
d
o
m
f
o
r
est
(
R
F)
,
lo
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
,
K
-
m
ea
n
s
c
l
u
s
ter
in
g
,
Gau
s
s
ian
Naiv
e
B
ay
es,
K
-
n
ea
r
est
n
eig
h
b
o
r
s
(
KNNs)
,
SVM,
d
ec
is
io
n
tr
ee
s
(
DT
)
,
C
NN,
R
NN,
C
NN+
R
NN,
E
S,
an
d
DANI
N.
T
h
e
ev
alu
atio
n
aim
s
to
co
m
p
ar
e
t
h
e
p
e
r
f
o
r
m
an
ce
o
f
th
ese
m
o
d
els
to
d
eter
m
in
e
th
e
m
o
s
t
ef
f
ec
tiv
e
m
eth
o
d
s
in
ter
m
s
o
f
th
ese
m
et
r
ics.
T
h
e
r
esu
lts
ar
e
s
h
o
wn
in
t
h
e
f
o
r
m
o
f
g
r
ap
h
s
an
d
tab
les.
4
.
1
.
Resul
t
s
Fig
u
r
e
2
d
ep
icts
th
e
AC
C
o
f
v
ar
io
u
s
ML
an
d
DL
m
o
d
els.
T
h
e
m
o
d
els
in
clu
d
e
RF
,
LR
,
K
-
m
ea
n
s
cl
u
s
ter
in
g
,
Gau
s
s
ian
Naiv
e
B
ay
es,
KNNs
,
SV
M,
DT
,
C
NN
,
R
NN,
C
NN+
R
NN,
E
S,
an
d
DANI
N.
T
h
e
ch
ar
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
2
,
Au
g
u
s
t
20
25
:
1
1
6
4
-
1
175
1170
r
ev
ea
ls
th
at
th
e
DANI
N
an
d
E
S
m
o
d
els
ex
h
ib
it
th
e
h
ig
h
est
a
cc
u
r
ac
y
,
clo
s
ely
f
o
llo
wed
b
y
C
NN+
R
NN,
R
NN,
an
d
C
NN
m
o
d
els,
all
o
f
wh
ic
h
s
u
r
p
ass
th
e
9
5
%
m
ar
k
.
DT
an
d
SVM
s
h
o
w
m
o
d
er
ately
h
i
g
h
ac
cu
r
ac
y
,
ab
o
v
e
8
0
%.
I
n
co
n
tr
ast,
KNNs
,
Gau
s
s
ian
Naiv
e
B
ay
es,
an
d
K
-
m
e
an
s
c
lu
s
ter
in
g
p
er
f
o
r
m
b
elo
w
8
0
%,
with
LR
an
d
R
F
h
av
in
g
th
e
lo
west
ac
cu
r
a
cies,
ar
o
u
n
d
7
0
%.
T
h
is
an
al
y
s
is
in
d
icate
s
a
clea
r
ad
v
a
n
t
ag
e
o
f
DL
m
o
d
els,
p
ar
ticu
lar
ly
th
o
s
e
th
at
co
m
b
i
n
e
C
NN
an
d
R
NN,
o
v
er
tr
ad
itio
n
al
ML
ap
p
r
o
ac
h
es in
ter
m
s
o
f
ac
cu
r
ac
y
.
Fig
u
r
e
2
.
Acc
u
r
ac
y
c
o
m
p
ar
is
o
n
o
f
e
x
is
tin
g
s
tate
-
of
-
ar
t te
ch
n
iq
u
es with
DANI
N
Fig
u
r
e
3
illu
s
tr
ates
th
e
p
r
ec
i
s
io
n
(
PR
E
)
o
f
v
ar
io
u
s
ML
a
n
d
DL
m
o
d
els,
in
cl
u
d
in
g
D
ANI
N,
E
S,
C
NN+
R
NN,
R
NN,
C
NN,
DT
,
SVM,
KNNs
,
Gau
s
s
ian
Naiv
e
B
ay
es,
K
-
m
ea
n
s
c
lu
s
ter
in
g
,
an
d
LR
.
T
h
e
DANI
N
an
d
E
S
m
o
d
els
ag
ai
n
to
p
th
e
ch
ar
t
with
t
h
e
h
ig
h
est
p
r
ec
is
io
n
v
alu
es,
i
n
d
icat
in
g
th
eir
s
u
p
e
r
io
r
p
er
f
o
r
m
an
ce
in
co
r
r
ec
tly
id
e
n
tify
in
g
p
o
s
itiv
e
in
s
tan
ce
s
.
T
h
e
C
NN+
R
NN,
R
NN,
an
d
C
NN
m
o
d
els
a
ls
o
d
em
o
n
s
tr
ate
h
i
g
h
p
r
ec
is
io
n
,
r
e
f
lectin
g
th
eir
ef
f
ec
tiv
e
n
ess
in
m
in
im
izin
g
f
alse
p
o
s
itiv
es.
DT
an
d
SVM
ex
h
ib
it
m
o
d
er
ately
h
ig
h
p
r
ec
is
io
n
,
f
allin
g
ju
s
t
b
elo
w
t
h
e
to
p
-
tie
r
m
o
d
els.
KNNs
,
Gau
s
s
ian
Naiv
e
B
ay
es,
an
d
K
-
m
ea
n
s
c
lu
s
ter
in
g
s
h
o
w
lo
wer
p
r
ec
is
io
n
,
in
d
icatin
g
a
h
i
g
h
er
r
ate
o
f
f
alse
p
o
s
itiv
es
co
m
p
ar
ed
t
o
th
e
to
p
-
p
er
f
o
r
m
in
g
m
o
d
els.
LR
,
wh
il
e
s
lig
h
tly
b
etter
th
an
K
-
m
ea
n
s
c
lu
s
ter
in
g
,
s
till
lag
s
b
eh
in
d
th
e
o
th
er
m
eth
o
d
s
.
T
h
is
an
aly
s
is
h
ig
h
lig
h
ts
th
e
s
u
p
er
io
r
p
r
ec
is
io
n
o
f
DL
m
o
d
els,
p
ar
ticu
lar
ly
t
h
e
co
m
b
in
ed
C
NN+
R
NN
ap
p
r
o
ac
h
,
co
m
p
ar
ed
to
tr
ad
itio
n
al
ML
m
o
d
els.
Fig
u
r
e
3
.
Pre
cisi
o
n
c
o
m
p
ar
is
o
n
o
f
e
x
is
tin
g
s
tate
-
of
-
ar
t te
ch
n
iq
u
es with
DANI
N
0
20
40
60
80
100
120
Ra
ndom
f
or
e
st
Lo
g
i
st
i
c
r
e
g
r
e
ss
i
o
n
K-
m
e
an
s
c
l
us
t
e
r
i
ng
G
au
ss
i
an
na
i
v
e
b
ay
e
s
K-
ne
ar
e
st
n
e
i
g
hb
o
r
s
S
V
M
D
e
c
i
si
o
n
t
r
e
e
s
C
N
N
RNN
C
N
N
+
R
N
N
ES
PS
Acc
u
ra
cy
Me
th
o
d
o
logy
A
C
C
0
20
40
60
80
100
120
Lo
g
i
st
i
c
r
e
g
r
e
ss
i
o
n
K-
m
e
an
s
c
l
us
t
e
r
i
ng
G
au
ss
i
an
na
i
v
e
b
ay
e
s
K-
ne
ar
e
st
n
e
i
g
hb
o
r
s
S
V
M
D
e
c
i
s
i
o
n
t
r
e
e
s
C
N
N
R
N
N
C
N
N
+
R
N
N
ES
PS
Prec
is
ion
Valu
e
Me
th
o
d
o
logy
P
R
E
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
d
va
n
ce
d
d
ee
p
a
tten
tio
n
n
eu
r
a
l in
feren
ce
n
etw
o
r
k
fo
r
en
h
a
n
ce
d
a
r
r
h
yth
mia
…
(
H.
S
u
mit
h
a
)
1171
Fig
u
r
e
4
p
r
esen
ts
th
e
(
R
E
)
o
f
v
ar
i
o
u
s
ML
an
d
DL
m
o
d
el
s
,
in
clu
d
in
g
DANI
N,
E
S,
C
NN+
R
NN,
R
NN,
C
NN,
DT
,
SVM,
KNN
s
,
Gau
s
s
ian
Naiv
e
B
ay
es,
K
-
m
ea
n
s
c
lu
s
ter
in
g
,
LR
,
an
d
RF
.
T
h
e
DANI
N
an
d
E
S
m
o
d
els
ex
h
ib
it
th
e
h
ig
h
est
r
ec
all
v
alu
es,
b
o
th
at
1
0
0
,
in
d
ica
tin
g
th
eir
ex
ce
p
tio
n
al
p
e
r
f
o
r
m
an
ce
in
id
en
tify
in
g
all
r
elev
an
t
in
s
tan
ce
s
(
tr
u
e
p
o
s
itiv
es).
T
h
e
C
NN+
R
NN
m
o
d
el
f
o
llo
ws
clo
s
ely
with
a
r
ec
all
v
alu
e
o
f
9
3
.
7
6
,
d
em
o
n
s
tr
atin
g
its
ef
f
ec
tiv
en
es
s
in
m
in
im
izin
g
f
alse
n
eg
ativ
es.
R
NN
an
d
C
NN
m
o
d
els
also
s
h
o
w
h
ig
h
r
ec
all
v
alu
es,
at
8
7
.
2
5
an
d
8
8
.
1
2
r
esp
ec
tiv
ely
,
s
h
o
wca
s
in
g
th
eir
ab
ilit
y
to
id
en
tify
a
lar
g
e
p
o
r
tio
n
o
f
r
elev
a
n
t
in
s
tan
ce
s
.
DT
an
d
SVM
s
h
o
w
m
o
d
er
atel
y
h
i
g
h
r
ec
all
v
al
u
es,
at
8
7
.
5
5
an
d
8
6
.
2
8
r
esp
e
ctiv
ely
,
in
d
icatin
g
a
g
o
o
d
b
alan
ce
b
etwe
en
tr
u
e
p
o
s
itiv
es
an
d
f
alse
n
e
g
ativ
es.
I
n
co
n
tr
ast,
KNNs
(
7
8
.
2
2
)
,
Gau
s
s
ian
Naiv
e
B
ay
es
(
7
9
.
4
4
)
,
a
n
d
K
-
m
ea
n
s
c
lu
s
ter
i
n
g
(
7
0
.
3
3
)
ex
h
i
b
it
lo
wer
r
ec
all
v
alu
es,
r
ef
lectin
g
a
h
i
g
h
er
r
at
e
o
f
f
alse
n
eg
ativ
es
co
m
p
ar
ed
to
th
e
to
p
-
p
e
r
f
o
r
m
in
g
m
o
d
els.
LR
an
d
R
F
p
er
f
o
r
m
s
im
ilar
ly
,
with
r
ec
all
v
alu
es
o
f
9
1
.
0
8
a
n
d
8
4
.
8
5
r
esp
ec
tiv
ely
,
s
lig
h
tly
b
elo
w
t
h
o
s
e
o
f
th
e
b
etter
-
p
e
r
f
o
r
m
in
g
m
o
d
els.
T
h
is
an
aly
s
is
u
n
d
e
r
s
co
r
es
th
e
s
u
p
er
io
r
r
ec
all
o
f
DL
m
o
d
els,
p
ar
ticu
l
ar
ly
th
e
c
o
m
b
in
e
d
C
NN+
R
N
N
ap
p
r
o
ac
h
,
c
o
m
p
ar
e
d
to
tr
a
d
itio
n
al
ML
m
o
d
els,
in
d
icatin
g
th
eir
r
o
b
u
s
tn
ess
in
id
en
tify
in
g
r
elev
an
t i
n
s
tan
ce
s
.
Fig
u
r
e
4
.
R
ec
all
co
m
p
a
r
is
o
n
o
f
ex
is
tin
g
s
tate
-
of
-
ar
t te
ch
n
iq
u
es with
DANI
N
Fig
u
r
e
5
s
h
o
ws
th
e
F1
-
s
co
r
e
o
f
v
ar
i
o
u
s
ML
an
d
DL
m
o
d
e
ls
,
in
clu
d
in
g
DANI
N,
E
S,
C
NN+
R
NN
,
R
NN,
C
NN,
DT
,
SVM,
KNN
s
,
Gau
s
s
ian
Naiv
e
B
ay
es,
K
-
m
ea
n
s
c
lu
s
ter
in
g
,
LR
,
an
d
RF
.
T
h
e
DANI
N
an
d
E
S
m
o
d
els
ac
h
iev
e
th
e
h
ig
h
est
F1
-
s
co
r
e
s
,
b
o
th
at
1
0
0
,
in
d
i
ca
tin
g
th
eir
b
alan
ce
d
p
er
f
o
r
m
an
ce
in
ter
m
s
o
f
p
r
ec
is
io
n
an
d
r
ec
all.
T
h
e
C
NN+
R
NN
m
o
d
el
f
o
llo
ws
clo
s
e
ly
with
an
F1
-
s
co
r
e
o
f
9
4
.
2
5
,
d
em
o
n
s
tr
atin
g
its
ef
f
ec
tiv
e
h
an
d
lin
g
o
f
b
o
th
tr
u
e
p
o
s
itiv
es
an
d
m
in
im
izin
g
f
alse
p
o
s
itiv
es
an
d
n
eg
ativ
es.
R
NN
an
d
C
N
N
m
o
d
els
also
p
e
r
f
o
r
m
well
with
F1
-
s
co
r
e
s
o
f
9
0
.
5
6
an
d
8
9
.
4
6
,
r
esp
ec
tiv
ely
,
s
h
o
w
ca
s
in
g
th
eir
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
.
DT
an
d
SVM
ex
h
ib
it
m
o
d
e
r
ately
h
ig
h
F1
-
s
co
r
e
s
,
at
8
5
.
8
1
,
an
d
r
ec
all.
I
n
co
n
tr
ast,
KNNs
(
7
8
.
8
6
)
,
Gau
s
s
ian
Naiv
e
B
ay
es
(
7
4
.
9
5
)
,
an
d
K
-
m
ea
n
s
c
lu
s
ter
in
g
(
7
6
.
5
7
)
s
h
o
w
lo
we
r
F1
-
s
co
r
e
s
,
in
d
icatin
g
less
b
alan
ce
d
p
er
f
o
r
m
an
ce
co
m
p
ar
ed
to
t
h
e
to
p
-
p
er
f
o
r
m
in
g
m
o
d
els.
LR
an
d
R
F
h
av
e
th
e
l
o
west
F1
-
s
co
r
e
s
,
at
7
8
.
6
5
an
d
7
4
.
7
2
,
r
esp
ec
tiv
el
y
,
in
d
icatin
g
th
eir
lo
wer
o
v
er
all
p
er
f
o
r
m
an
ce
.
T
h
is
an
al
y
s
is
h
ig
h
lig
h
ts
th
e
s
u
p
er
io
r
F1
-
s
co
r
e
s
o
f
DL
m
o
d
els,
p
ar
ticu
lar
ly
th
e
c
o
m
b
in
e
d
C
NN+
R
N
N
ap
p
r
o
ac
h
,
c
o
m
p
ar
ed
to
tr
a
d
itio
n
al
ML
m
o
d
els.
T
ab
le
2
p
r
esen
ts
a
tab
u
lar
c
o
m
p
ar
is
o
n
o
f
v
ar
io
u
s
m
o
d
els
ac
r
o
s
s
m
u
ltip
le
m
etr
ics,
in
clu
d
in
g
r
ec
all
(
R
E
)
an
d
p
r
ec
is
io
n
(
PR
E
)
f
o
r
d
if
f
er
en
t
class
es
(
N,
S,
V,
F)
.
Fig
u
r
e
6
illu
s
tr
ates
th
e
p
er
f
o
r
m
an
ce
o
f
v
ar
io
u
s
m
o
d
els
ac
r
o
s
s
m
u
ltip
le
m
etr
ic
s
,
in
clu
d
in
g
r
ec
all
(
R
E
)
a
n
d
p
r
ec
is
io
n
(
PR
E
)
f
o
r
d
if
f
e
r
en
t
cl
ass
es
(
N,
S,
V,
F).
T
h
e
m
o
d
els
co
m
p
ar
ed
a
r
e
M
R
FO
-
SV
M,
f
u
s
ed
tr
an
s
f
o
r
m
e
r
,
DNN
-
en
s
em
b
le,
lig
h
t
tr
an
s
f
o
r
m
er
,
W
av
eI
Net
-
Db
6
,
W
av
eI
Net
-
Sy
m
4
,
MA
HA,
I
SF
n
et
-
1
4
,
E
C
GT
r
an
s
Fo
r
m
,
E
S,
a
n
d
DANI
N.
D
A
NI
N
an
d
E
S
m
o
d
els
co
n
s
is
ten
tly
ac
h
iev
e
h
ig
h
v
al
u
es
ac
r
o
s
s
all
m
etr
ics,
in
d
ica
tin
g
th
eir
r
o
b
u
s
t
p
e
r
f
o
r
m
an
ce
with
n
ea
r
-
p
e
r
f
ec
t
r
ec
all
an
d
p
r
ec
is
io
n
v
al
u
es.
E
C
GT
r
an
s
Fo
r
m
an
d
MA
HA
als
o
d
em
o
n
s
tr
ate
s
tr
o
n
g
p
er
f
o
r
m
an
ce
,
p
ar
ticu
lar
ly
in
r
ec
all
an
d
p
r
ec
is
io
n
f
o
r
m
o
s
t
class
es,
p
o
s
itio
n
in
g
th
em
am
o
n
g
th
e
to
p
m
o
d
els.
I
n
co
n
tr
ast,
I
SF
n
et
-
1
4
ex
h
ib
its
a
s
ig
n
if
ican
t
d
r
o
p
in
s
o
m
e
m
etr
ics,
p
ar
ticu
lar
ly
in
r
ec
all
(
N)
an
d
p
r
ec
is
io
n
(
N)
,
s
u
g
g
est
in
g
a
h
ig
h
e
r
r
ate
o
f
0
20
40
60
80
100
120
R
an
do
m
f
o
r
e
st
Lo
g
i
st
i
c
r
e
g
r
e
ss
i
o
n
K-
m
e
an
s
c
l
us
t
e
r
i
ng
G
au
ss
i
an
na
i
v
e
b
ay
e
s
K-
ne
ar
e
st
n
e
i
g
hb
o
r
s
S
V
M
D
e
c
i
si
o
n
t
r
e
e
s
C
N
N
R
N
N
C
N
N
+
R
N
N
ES
PS
Re
aca
ll
Valu
e
Me
th
o
d
o
logy
RE
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
2
,
Au
g
u
s
t
20
25
:
1
1
6
4
-
1
175
1172
f
alse
n
eg
ativ
es
an
d
less
ac
cu
r
ate
p
o
s
itiv
e
p
r
e
d
ictio
n
s
f
o
r
th
i
s
class
.
L
ig
h
t
tr
an
s
f
o
r
m
er
an
d
f
u
s
ed
tr
a
n
s
f
o
r
m
er
m
ain
tain
b
alan
ce
d
p
er
f
o
r
m
an
ce
ac
r
o
s
s
m
o
s
t
m
etr
ics,
th
o
u
g
h
n
o
t
as
h
ig
h
as
th
e
to
p
-
p
er
f
o
r
m
in
g
DANI
N
an
d
E
S
m
o
d
els.
W
av
eI
Net
-
Db
6
an
d
W
av
eI
Net
-
Sy
m
4
s
h
o
w
v
ar
y
in
g
r
esu
lts
,
with
n
o
tab
le
d
r
o
p
s
in
ce
r
tain
m
etr
ics,
re
f
lectin
g
less
co
n
s
is
ten
cy
.
DNN
en
s
em
b
le
d
is
p
lay
s
lo
wer
p
er
f
o
r
m
an
ce
,
esp
ec
ially
in
r
ec
all
(
V)
an
d
p
r
ec
is
io
n
(
V)
,
in
d
icatin
g
ch
alle
n
g
es
in
c
o
r
r
ec
tly
id
en
tif
y
in
g
an
d
class
if
y
in
g
r
elev
a
n
t
in
s
tan
ce
s
in
th
ese
class
e
s
.
Ov
er
all,
th
e
ch
ar
t
h
ig
h
lig
h
ts
th
e
s
u
p
er
io
r
a
n
d
c
o
n
s
is
ten
t
p
er
f
o
r
m
an
c
e
o
f
DANI
N
an
d
E
S
m
o
d
els
ac
r
o
s
s
all
m
etr
ics,
wh
ile
o
th
er
m
o
d
els
s
h
o
w
v
a
r
y
in
g
d
eg
r
ee
s
o
f
ef
f
ec
tiv
en
ess
,
with
s
o
m
e
ex
ce
llin
g
in
s
p
ec
if
ic
ar
ea
s
b
u
t
n
o
t
ac
r
o
s
s
th
e
b
o
ar
d
.
Fig
u
r
e
5
.
F1
-
s
co
r
e
co
m
p
ar
is
o
n
o
f
ex
is
tin
g
s
tate
-
of
-
ar
t te
ch
n
i
q
u
es with
DANI
N
T
ab
le
2
.
C
o
m
p
a
r
is
o
n
tab
le
M
e
t
h
o
d
s
R
E
(N)
P
R
E
(N)
R
E
(S)
P
R
E
(S)
R
E
(V)
P
R
E
(V)
R
E
(F)
P
R
E
(F)
A
v
e
r
a
g
e
RE
A
v
e
r
a
g
e
P
R
E
A
v
e
r
a
g
e
F1
-
sco
r
e
M
R
F
O
-
S
V
M
[
1
6
]
9
8
.
8
9
8
.
7
98
9
9
.
3
9
8
.
2
9
6
.
4
96
9
7
.
8
9
7
.
8
9
7
.
6
9
7
.
7
9
7
.
7
F
u
se
d
t
r
a
n
sf
o
r
mer [
1
7
]
9
9
.
2
9
9
.
2
9
4
.
8
9
1
.
4
6
9
.
3
8
7
.
4
8
6
.
5
8
7
.
7
9
2
.
6
9
0
.
1
9
0
.
1
9
0
.
1
DNN
-
e
n
sem
b
l
e
[
1
8
]
9
3
.
1
9
8
.
1
8
3
.
1
9
0
.
9
4
9
.
4
8
0
.
2
4
9
.
4
7
6
.
4
6
8
.
7
8
6
.
4
78
78
Li
g
h
t
t
r
a
n
sf
o
r
mer
[
1
9
]
9
9
.
8
9
9
.
7
9
1
.
5
9
4
.
4
83
9
3
.
8
9
1
.
4
9
1
.
4
9
1
.
4
9
3
.
8
9
4
.
8
9
3
.
6
W
a
v
e
I
N
e
t
-
D
b
6
[
2
0
]
92
9
5
.
3
7
4
.
6
6
4
.
8
6
4
.
8
6
4
.
8
6
4
.
8
6
4
.
8
7
4
.
1
6
4
.
8
6
1
.
3
6
1
.
3
W
a
v
e
I
N
e
t
-
S
y
m4
[
2
1
]
9
1
.
4
9
7
.
7
9
1
.
4
6
5
.
7
4
9
.
3
2
5
.
6
6
4
.
8
7
7
.
3
6
3
.
8
6
6
.
6
6
9
.
4
6
9
.
4
M
A
H
A
[
2
2
]
9
9
.
5
9
1
.
9
9
9
.
4
85
8
4
.
9
9
0
.
1
9
1
.
8
9
1
.
8
9
3
.
9
8
9
.
7
8
9
.
7
9
3
.
2
I
S
F
n
e
t
-
1
4
[
2
3
]
6
3
.
5
6
3
.
8
9
6
.
1
9
1
.
8
9
9
.
4
9
8
.
6
8
5
.
4
8
3
.
1
8
6
.
1
8
4
.
3
8
4
.
2
8
4
.
2
EC
G
Tr
a
n
sF
o
r
m
[
2
4
]
9
9
.
7
9
7
.
8
9
9
.
7
8
9
.
2
9
3
.
8
8
6
.
5
9
6
.
6
9
3
.
6
9
7
.
6
9
1
.
8
9
3
.
3
95
ES [
2
5
]
9
9
.
8
9
9
.
7
9
7
.
5
9
7
.
7
9
1
.
7
9
5
.
7
1
0
0
9
6
.
8
9
7
.
5
9
5
.
2
9
6
.
4
9
6
.
4
D
A
N
I
N
(
P
S
)
9
9
.
9
9
9
.
8
9
8
.
4
9
8
.
8
7
9
3
.
4
3
9
6
.
7
6
1
0
0
9
7
.
8
6
9
8
.
7
6
9
6
.
7
6
9
7
.
8
6
9
7
.
8
7
Fig
u
r
e
7
p
r
esen
ts
an
av
er
a
g
e
co
m
p
ar
is
o
n
o
f
v
ar
io
u
s
m
o
d
els
b
ased
o
n
th
r
ee
m
etr
ics:
av
e
r
ag
e
r
ec
all
(
R
E
)
,
av
er
ag
e
p
r
ec
is
io
n
(
PR
E
)
,
an
d
a
v
e
r
ag
e
F1
-
s
co
r
e.
T
h
e
m
o
d
els
co
m
p
ar
ed
in
clu
d
e
MRF
O
-
S
VM
,
f
u
s
ed
tr
an
s
f
o
r
m
er
,
DNN
-
en
s
em
b
le,
lig
h
t
tr
an
s
f
o
r
m
er
,
W
av
eI
Net
-
Db
6
,
W
a
v
eI
Net
-
Sy
m
4
,
MA
HA,
I
SF
n
et
-
1
4
,
E
C
GT
r
an
s
Fo
r
m
,
E
S,
an
d
D
ANI
N.
Fro
m
th
e
ch
ar
t,
it
is
ev
id
en
t
th
at
th
e
DANI
N
an
d
E
S
m
o
d
els
ex
h
ib
it
s
u
p
er
io
r
p
e
r
f
o
r
m
an
ce
ac
r
o
s
s
all
th
r
ee
m
etr
ics,
with
v
alu
es
cl
o
s
e
to
o
r
r
ea
c
h
in
g
1
0
0
,
in
d
ica
tin
g
th
eir
ex
ce
llen
t
b
alan
ce
b
etwe
en
r
e
ca
ll,
p
r
ec
is
io
n
,
an
d
o
v
er
all
ef
f
ec
tiv
en
ess
.
E
C
GT
r
an
s
Fo
r
m
an
d
MA
HA
also
p
er
f
o
r
m
well
,
s
h
o
win
g
h
ig
h
av
er
a
g
e
v
alu
es a
cr
o
s
s
all
m
etr
ics,
p
lacin
g
th
em
am
o
n
g
th
e
to
p
p
er
f
o
r
m
er
s
.
0
20
40
60
80
100
120
R
an
do
m
f
o
r
e
st
Lo
g
i
st
i
c
r
e
g
r
e
ss
i
o
n
K-
m
e
an
s
c
l
us
t
e
r
i
ng
G
au
ss
i
an
na
i
v
e
b
ay
e
s
K-
ne
ar
e
st
n
e
i
g
hb
o
r
s
S
V
M
D
e
c
i
si
o
n
t
r
e
e
s
C
N
N
R
N
N
C
N
N
+
R
N
N
ES
PS
F1
-
s
core
Me
th
o
d
o
logy
F1
-
sc
or
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
A
d
va
n
ce
d
d
ee
p
a
tten
tio
n
n
eu
r
a
l in
feren
ce
n
etw
o
r
k
fo
r
en
h
a
n
ce
d
a
r
r
h
yth
mia
…
(
H.
S
u
mit
h
a
)
1173
Fig
u
r
e
6
.
N,
S,
V,
F c
o
m
p
ar
is
o
n
o
f
ex
is
tin
g
s
tate
-
of
-
th
e
-
ar
t t
ec
h
n
iq
u
es with
DANI
N
L
ig
h
t
tr
an
s
f
o
r
m
er
a
n
d
f
u
s
ed
tr
an
s
f
o
r
m
er
d
e
m
o
n
s
tr
ate
b
ala
n
ce
d
p
er
f
o
r
m
an
ce
,
with
av
er
ag
e
v
alu
es
co
n
s
is
ten
tly
h
ig
h
,
alth
o
u
g
h
s
lig
h
tly
lo
wer
th
a
n
t
h
e
to
p
m
o
d
els
DANI
N
a
n
d
E
S.
W
av
eI
Net
-
Db
6
an
d
W
av
eI
Net
-
Sy
m
4
s
h
o
w
m
o
r
e
v
ar
iab
ilit
y
,
with
s
ig
n
if
ican
t
d
r
o
p
s
in
s
o
m
e
m
etr
ics,
r
ef
lect
in
g
less
co
n
s
is
ten
t
p
er
f
o
r
m
an
ce
.
DNN
-
en
s
em
b
le
an
d
I
SF
n
et
-
1
4
h
av
e
n
o
ticea
b
l
y
lo
wer
av
er
a
g
e
v
alu
es,
p
ar
tic
u
lar
ly
in
th
e
r
ec
all,
in
d
icatin
g
ch
allen
g
es
in
id
en
tify
in
g
a
ll
r
elev
a
n
t
in
s
tan
ce
s
.
T
h
is
lo
wer
p
er
f
o
r
m
an
ce
im
p
ac
ts
th
eir
o
v
er
all
F1
-
s
co
r
e
,
s
u
g
g
esti
n
g
a
n
ee
d
f
o
r
im
p
r
o
v
em
e
n
t
in
t
h
ese
m
o
d
els.
Ov
er
all,
th
e
ch
a
r
t
h
ig
h
lig
h
ts
th
e
ex
ce
p
tio
n
al
an
d
c
o
n
s
is
ten
t
p
er
f
o
r
m
a
n
ce
o
f
DANI
N
an
d
E
S
m
o
d
els
ac
r
o
s
s
all
m
etr
ics.
Oth
er
m
o
d
els
li
k
e
E
C
GT
r
an
s
Fo
r
m
an
d
MA
HA
also
s
h
o
w
s
tr
o
n
g
p
er
f
o
r
m
an
ce
,
wh
ile
s
o
m
e
m
o
d
els
lik
e
DNN
-
en
s
em
b
le
an
d
I
SF
n
et
-
1
4
r
eq
u
ir
e
f
u
r
th
er
o
p
tim
izatio
n
t
o
im
p
r
o
v
e
th
eir
av
er
a
g
e
r
ec
all
an
d
p
r
ec
is
io
n
.
Fig
u
r
e
7
.
Av
e
r
ag
e
co
m
p
ar
is
o
n
o
f
ex
is
tin
g
s
tate
-
of
-
ar
t te
ch
n
i
q
u
es with
DANI
N
5.
CO
NCLU
SI
O
N
I
n
co
n
clu
s
io
n
,
t
h
e
DANI
N
m
eth
o
d
o
l
o
g
y
s
ig
n
if
ican
tly
e
n
h
an
ce
s
th
e
ac
c
u
r
ac
y
an
d
tim
elin
ess
o
f
ar
r
h
y
th
m
ia
d
etec
tio
n
b
y
in
te
g
r
atin
g
o
n
e
-
d
im
e
n
s
io
n
al
E
C
G
s
ig
n
als
with
two
-
d
im
en
s
io
n
al
s
p
ec
tr
al
im
ag
es.
T
h
is
in
n
o
v
ativ
e
ap
p
r
o
ac
h
lev
e
r
ag
es
d
ee
p
atten
tio
n
n
etwo
r
k
-
b
ased
m
o
d
els
f
o
r
s
u
p
er
io
r
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
in
co
r
p
o
r
ates
an
in
f
er
e
n
ce
m
o
d
el
s
y
s
tem
f
o
r
im
p
r
o
v
ed
in
te
r
p
r
etab
ilit
y
an
d
clin
ical
u
s
ab
i
lity
.
C
o
m
p
ar
ativ
e
an
aly
s
is
with
tr
ad
itio
n
al
an
d
s
tate
-
of
-
th
e
-
ar
t
m
eth
o
d
s
d
em
o
n
s
tr
ates
th
at
DANI
N
ac
h
iev
es
h
ig
h
er
p
er
f
o
r
m
an
ce
0
20
40
60
80
100
120
v
alu
e
Me
th
od
N
,S,
V
,F
c
omp
aris
on
R
E
(
N
)
PR
E
(
N
)
R
E
(
S
)
PR
E
(
S
)
R
E
(
V
)
PR
E
(
V
)
R
E
(
F
)
PR
E
(
F
)
0
20
40
60
80
100
120
v
alu
e
Me
th
o
d
A
v
er
ag
e
c
omp
aris
on
A
v
e
rage R
E
A
ve
rag
e
P
R
E
A
ve
rag
e
F
1 S
c
o
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.