I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
7
,
No
.
2
,
J
u
n
e
201
8
,
p
p
.
90
~
94
I
SS
N:
2252
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
7
.
i2
.
p
p
90
-
94
90
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e.
co
m/jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JA
I
Cla
ss
ificatio
n of
Atrial Ar
rhy
th
m
i
a
s usin
g
Neural N
etw
o
rk
s
J
a
i U
t
k
a
rsh
1
,
Ra
j
u K
u
m
a
r
P
a
nd
ey
2
,
Sh
re
y
K
u
m
a
r
D
ub
ey
3
,
Sh
ub
ha
m
Sin
ha
4
,
S.
S.
Sa
hu
5
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
C
o
m
m
u
n
ica
ti
o
n
E
n
g
in
e
e
rin
g
,
Birl
a
In
stit
u
te o
f
T
e
c
h
n
o
lo
g
y
,
M
e
sra
,
In
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
2
1
,
2
0
1
8
R
ev
i
s
ed
Feb
7
,
2
0
1
8
A
cc
ep
ted
Ma
y
19
,
2
0
1
8
El
e
c
tro
c
a
rd
io
g
ra
m
(EC
G
)
is
a
n
im
p
o
rtan
t
t
o
o
l
u
se
d
b
y
c
li
n
icia
n
s
f
o
r
su
c
c
e
ss
f
u
l
d
iag
n
o
sis
a
n
d
d
e
tec
ti
o
n
o
f
A
rrh
y
th
m
ia
s,
li
k
e
A
tri
a
l
F
ib
ril
lati
o
n
(A
F
)
a
n
d
A
tri
a
l
F
lu
tt
e
r
(A
F
L
).
In
th
is
m
a
n
u
sc
rip
t,
a
n
e
f
f
icie
n
t
te
c
h
n
iq
u
e
o
f
c
las
si
fy
in
g
a
tri
a
l
a
rrh
y
th
m
ias
f
ro
m
No
r
m
a
l
S
in
u
s
Rh
y
th
m
(NSR)
h
a
s
b
e
e
n
p
re
se
n
ted
.
A
u
to
re
g
re
ss
iv
e
M
o
d
e
ll
in
g
h
a
s
b
e
e
n
u
se
d
t
o
c
a
p
t
u
re
th
e
f
e
a
tu
re
s
o
f
th
e
ECG
sig
n
a
l,
w
h
ich
a
re
th
e
n
f
e
d
a
s
in
p
u
ts
to
th
e
n
e
u
ra
l
n
e
tw
o
rk
f
o
r
c
las
si
f
ica
ti
o
n
.
T
h
e
sta
n
d
a
rd
d
a
ta
b
a
se
a
v
a
il
a
b
le
a
t
P
h
y
sio
n
e
t
Ba
n
k
re
p
o
sit
o
ry
h
a
s
b
e
e
n
u
se
d
f
o
r
train
in
g
,
v
a
li
d
a
ti
o
n
a
n
d
tes
ti
n
g
o
f
th
e
m
o
d
e
l.
Ex
h
a
u
stiv
e
e
x
p
e
ri
m
e
n
tal
stu
d
y
h
a
s
b
e
e
n
c
a
rried
o
u
t
b
y
e
x
tra
c
ti
n
g
EC
G
s
a
m
p
les
o
f
d
u
ra
ti
o
n
o
f
5
se
c
o
n
d
s,
1
0
se
c
o
n
d
s
a
n
d
2
0
se
c
o
n
d
s.
I
t
p
r
o
v
id
e
s
a
n
a
c
c
u
ra
c
y
o
f
9
9
%
a
n
d
9
4
.
3
%
o
n
train
in
g
a
n
d
t
e
st
se
t
re
sp
e
c
ti
v
e
l
y
f
o
r
5
se
c
re
c
o
rd
in
g
s
.
I
n
1
0
se
c
a
n
d
2
0
se
c
sa
m
p
le
s
it
sh
o
w
s
1
0
0
%
a
c
c
u
ra
c
y
.
T
h
u
s,
th
e
p
ro
p
o
se
d
m
e
th
o
d
c
a
n
b
e
u
se
d
t
o
d
e
tec
t
t
h
e
a
rrh
y
th
m
ias
in
a
s
m
a
ll
d
u
ra
ti
o
n
re
c
o
rd
in
g
s
w
it
h
a
f
a
irl
y
h
ig
h
a
c
c
u
ra
c
y
.
K
ey
w
o
r
d
:
A
tr
ial
Fib
r
illatio
n
A
tr
ial
Flu
tter
Au
to
r
eg
r
es
s
i
v
e
Mo
d
ellin
g
E
C
G
Neu
r
al
Net
w
o
r
k
s
Co
p
y
rig
h
t
©
2
0
1
8
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
S.
S.
Sah
u
,
Dep
ar
t
m
en
t o
f
E
lectr
o
n
ics a
n
d
C
o
m
m
u
n
icat
io
n
E
n
g
i
n
ee
r
in
g
,
B
ir
la
I
n
s
titu
te
Of
T
ec
h
n
o
lo
g
y
,
Me
s
r
a,
R
an
c
h
i,
J
h
ar
k
h
a
n
d
,
8
3
5
2
1
5
,
I
n
d
ia
.
E
m
ail: s
s
s
a
h
u
@
b
it
m
e
s
r
a.
ac
.
in
1.
I
NT
RO
D
UCT
I
O
N
E
lectr
o
ca
r
d
io
g
r
am
o
r
E
C
G
i
s
u
s
ed
to
r
ec
o
r
d
th
e
elec
tr
ical
ac
tiv
it
y
o
f
t
h
e
h
ea
r
t
o
v
er
a
ce
r
tain
ti
m
e
p
er
io
d
.
A
n
y
ab
n
o
r
m
alit
y
o
r
d
ev
iatio
n
f
r
o
m
n
o
r
m
al
E
C
G
ca
n
p
r
o
v
e
to
b
e
f
a
tal
f
o
r
p
ati
en
ts
.
Mo
s
tl
y
,
t
h
ese
ab
n
o
r
m
alitie
s
ar
e
o
b
s
er
v
ed
b
y
t
h
e
n
a
k
ed
e
y
e
s
o
f
t
h
e
d
o
cto
r
an
d
h
av
e
co
n
s
id
er
ab
le
ch
an
ce
s
o
f
b
ein
g
u
n
d
etec
ted
.
T
w
o
o
f
th
e
m
o
s
t
co
m
m
o
n
lif
e
t
h
r
ea
ten
in
g
A
r
r
h
y
th
m
ia
s
ar
e
A
tr
ial
Fib
r
illa
tio
n
an
d
A
tr
ial
Fl
u
tter
[1
-
6
]
.
A
tr
ial
f
ib
r
illatio
n
in
v
o
l
v
es
ir
r
e
g
u
lar
h
ea
r
t
b
ea
t.
I
t
h
a
s
n
o
v
i
s
ib
le
s
y
m
p
to
m
s
an
d
i
s
ter
m
ed
as
a
‘
Si
len
t
Dis
ea
s
e
’
,
m
a
k
i
n
g
its
d
iag
n
o
s
i
s
ex
tr
e
m
el
y
d
i
f
f
icu
l
t.
I
f
r
e
m
ain
u
n
d
etec
ted
,
it
ca
n
lead
to
s
tr
o
k
e,
h
ea
r
t
f
ail
u
r
e
o
r
ev
en
s
u
d
d
en
d
ea
t
h
s
[
7
]
.
Stat
is
tics
s
u
g
g
e
s
t
t
h
at
ap
p
r
o
x
i
m
a
tel
y
3
3
.
7
m
illi
o
n
p
eo
p
le
w
o
r
ld
w
id
e
h
a
v
e
atr
ial
f
ib
r
illatio
n
,
w
ith
th
e
m
aj
o
r
ity
co
n
s
tit
u
ti
n
g
d
e
v
elo
p
ed
n
ati
o
n
s
[
8
]
.
A
tr
ial
f
l
u
tter
i
s
th
e
ab
n
o
r
m
alit
y
i
n
th
e
b
ea
tin
g
r
ate
o
f
th
e
h
ea
r
t
d
u
e
t
o
an
ab
n
o
r
m
al
co
n
d
u
ct
io
n
cir
cu
it
d
ev
elo
p
in
g
in
s
id
e
r
ig
h
t
at
r
iu
m
.
T
h
is
r
ed
u
ce
s
th
e
b
lo
o
d
p
u
m
p
i
n
g
ca
p
ac
it
y
o
f
th
e
h
ea
r
t
a
n
d
ca
n
lead
to
m
aj
o
r
h
ea
r
t
d
is
ea
s
es
lik
e
s
tr
o
k
e,
co
n
g
esti
v
e
h
ea
r
t
f
ail
u
r
e
an
d
h
ea
r
t
attac
k
[
9
]
.
Var
io
u
s
s
i
g
n
al
p
r
o
ce
s
s
i
n
g
tech
n
iq
u
es
h
a
v
e
b
ee
n
u
s
ed
in
an
att
e
m
p
t
to
d
etec
t
an
d
class
i
f
y
atr
ial
f
ib
r
illatio
n
a
n
d
atr
ial
f
l
u
tter
.
T
i
m
e
F
r
eq
u
en
c
y
an
al
y
s
is
,
R
R
i
n
ter
v
a
l ir
r
eg
u
lar
it
y
,
Hid
d
en
Ma
r
k
o
v
Mo
d
els ar
e
f
e
w
o
f
th
e
p
ast tec
h
n
iq
u
es
u
s
ed
f
o
r
s
ep
ar
atio
n
o
f
A
F a
n
d
A
F
L
f
r
o
m
n
o
r
m
a
l si
n
u
s
r
h
y
t
h
m
.
Ho
w
e
v
er
,
th
e
y
a
ll
s
u
f
f
er
ed
f
r
o
m
i
n
h
er
en
t
d
is
ad
v
an
tag
e
s
d
u
e
to
q
u
ick
c
h
an
g
i
n
g
r
h
y
th
m
s
o
f
E
C
G
s
ig
n
al
s
[
1
0
-
1
5
]
.
Dif
f
er
e
n
t
tech
n
iq
u
es
h
a
v
e
b
ee
n
u
s
ed
in
an
atte
m
p
t
to
class
i
f
y
t
h
e
E
C
G
s
i
g
n
al
s
w
it
h
a
m
ax
i
m
u
m
p
o
s
s
ib
le
ac
c
u
r
ac
y
[
1
6
-
2
0
]
.
A
n
ac
cu
r
ac
y
o
f
9
3
%
h
av
e
b
ee
n
r
ep
o
r
ted
in
a
m
an
u
s
c
r
ip
t
u
s
i
n
g
a
f
ea
t
u
r
e
ex
tr
ac
tio
n
tech
n
iq
u
e
b
ased
o
n
t
h
e
p
r
o
d
u
ct
o
f
th
e
p
u
ls
e
d
u
r
atio
n
,
p
u
ls
e
ar
ea
an
d
p
u
ls
e
s
lo
p
e
o
f
th
e
f
il
ter
ed
a
m
b
u
lato
r
y
E
C
G
s
i
g
n
al
[
2
1
]
.
Fu
r
t
h
er
m
o
r
e,
a
m
an
u
s
cr
ip
t
u
s
i
n
g
A
r
ti
f
icial
Ne
u
r
al
Ne
t
w
o
r
k
(
A
NN)
,
r
ep
o
r
ts
an
ac
cu
r
ac
y
o
f
8
2
%
in
ar
r
h
y
t
h
m
i
a
class
i
f
icatio
n
[
2
2
]
.
An
o
th
er
p
ap
er
u
s
in
g
d
ee
p
C
o
n
v
o
l
u
tio
n
al
Neu
r
al
Ne
t
w
o
r
k
(
C
NN)
as
t
h
e
lear
n
in
g
al
g
o
r
it
h
m
r
ep
o
r
ts
an
ac
cu
r
ac
y
o
f
9
7
%
[
1
]
.
A
n
o
t
h
er
m
an
u
s
cr
ip
t
u
s
es
a
S
u
p
p
o
r
t
Vec
to
r
Ma
ch
i
n
e
(
SVM)
class
if
ier
to
r
ep
o
r
t a
n
ac
cu
r
ac
y
o
f
9
7
% [
7
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
C
la
s
s
i
fica
tio
n
o
f A
tr
ia
l A
r
r
h
yth
mia
s
u
s
in
g
N
eu
r
a
l Netw
o
r
k
s
(
Ja
i U
tka
r
s
h
)
91
I
n
th
i
s
m
an
u
s
cr
ip
t,
a
n
e
u
r
al
n
et
w
o
r
k
b
ased
ef
f
icie
n
t
m
o
d
el
h
as
b
ee
n
p
r
o
p
o
s
ed
to
i
m
p
r
o
v
e
th
e
d
etec
tio
n
an
d
cla
s
s
i
f
icatio
n
o
f
ar
r
h
y
t
h
m
ia
s
f
r
o
m
n
o
r
m
al
s
i
n
u
s
r
h
y
t
h
m
.
Firstl
y
,
t
h
e
f
ea
t
u
r
es
o
f
th
e
E
C
G
s
i
g
n
a
ls
ar
e
ex
tr
ac
ted
u
s
i
n
g
Au
to
r
eg
r
es
s
iv
e
m
o
d
elli
n
g
a
n
d
th
e
n
,
t
h
ese
f
ea
t
u
r
es a
r
e
f
ed
i
n
to
a
n
e
u
r
al
n
et
w
o
r
k
to
cla
s
s
i
f
y
th
e
ar
r
h
y
th
m
ia
s
.
T
h
is
p
ap
er
is
d
iv
id
ed
in
to
4
s
ec
tio
n
s
.
Sectio
n
2
g
i
v
es
t
h
e
m
et
h
o
d
o
lo
g
y
i
n
v
o
lv
ed
in
d
esi
g
n
in
g
th
e
e
f
f
ic
ien
t
c
lass
if
ica
tio
n
m
o
d
el
o
f
5
s
ec
,
1
0
s
ec
an
d
2
0
s
ec
d
u
r
atio
n
ar
r
h
y
th
m
ia
s
s
a
m
p
les.
Sect
io
n
3
d
escr
ib
es
th
e
ar
tific
ial
n
e
u
r
al
n
et
w
o
r
k
u
s
ed
to
o
b
tain
th
e
r
es
u
lts
.
Sec
tio
n
4
g
iv
e
s
t
h
e
f
i
n
al
r
esu
lt
s
,
f
o
llo
w
ed
b
y
th
e
o
v
er
all
co
n
cl
u
s
io
n
o
f
t
h
e
w
o
r
k
i
n
Sectio
n
5
.
2.
M
E
T
H
O
DO
L
O
G
Y
2
.
1
.
Da
t
a
ba
s
e
T
h
e
ex
is
ti
n
g
s
ta
n
d
ar
d
d
atab
ase
av
ailab
le
at
P
h
y
s
io
n
et
b
an
k
ar
ch
iv
e
h
a
s
b
ee
n
u
s
ed
f
o
r
tr
ain
i
n
g
an
d
test
i
n
g
o
f
t
h
e
m
o
d
el.
T
h
e
MI
T
/B
I
H
at
r
ial
f
ib
r
illatio
n
d
atab
ase
[
2
3
]
co
m
p
r
is
ed
o
f
2
3
atr
ial
f
lu
tter
E
C
G
r
ec
o
r
d
in
g
s
,
s
a
m
p
led
at
th
e
r
at
e
o
f
1
2
5
0
Hz.
T
h
e
MI
T
/B
I
H
No
r
m
a
l
s
i
n
u
s
r
h
y
t
h
m
d
atab
ase
[
2
4
]
h
ad
1
8
n
o
r
m
al
s
in
u
s
r
h
y
th
m
r
ec
o
r
d
ed
an
d
s
a
m
p
led
at
1
2
8
Hz.
T
h
e
MI
T
/
B
I
H
ar
r
h
y
t
h
m
ia
d
atab
ase
[
2
5
]
co
n
s
is
ted
o
f
3
r
ec
o
r
d
s
o
f
atr
ial
f
lu
t
ter
,
at
th
e
s
a
m
p
li
n
g
f
r
eq
u
e
n
c
y
o
f
3
6
0
Hz.
Ou
t
o
f
t
h
e
o
b
tain
ed
s
a
m
p
les,
th
e
tr
ai
n
i
n
g
a
n
d
test
i
n
g
o
f
t
h
e
s
a
m
p
les
h
av
e
b
e
en
d
o
n
e
b
y
e
x
tr
ac
ti
n
g
s
i
g
n
al
s
o
f
5
s
ec
d
u
r
atio
n
,
1
0
s
ec
d
u
r
at
io
n
an
d
2
0
s
ec
o
n
d
d
u
r
atio
n
.
T
h
e
n
u
m
b
er
o
f
tr
ai
n
in
g
a
n
d
tes
t
s
a
m
p
les
f
o
r
ea
ch
o
f
a
tr
ial
f
ib
r
illatio
n
,
f
l
u
tte
r
an
d
n
o
r
m
a
l
s
i
n
u
s
r
h
y
t
h
m
f
o
r
th
e
th
r
ee
s
a
m
p
le
d
u
r
atio
n
h
a
s
b
ee
n
p
r
esen
ted
in
T
ab
le
1
.
I
t
ca
n
b
e
s
ee
n
f
r
o
m
T
ab
le
1
th
at
th
e
to
tal
s
a
m
p
les ar
e
d
iv
id
ed
in
th
e
r
ati
o
o
f
4
:1
as tr
ain
in
g
s
et
an
d
te
s
t
s
et.
T
ab
le
1
.
Nu
m
b
er
o
f
s
a
m
p
le
s
f
o
r
ar
r
h
y
t
h
m
ias d
etec
tio
n
S
a
mp
l
e
D
u
r
a
t
i
o
n
A
t
r
i
a
l
F
i
b
r
i
l
l
a
t
i
o
n
A
t
r
i
a
l
F
l
u
t
t
e
r
N
o
r
mal
S
i
n
u
s
R
h
y
t
h
m
T
r
a
i
n
i
n
g
samp
l
e
s
T
e
st
i
n
g
S
a
mp
l
e
s
T
o
t
a
l
T
r
a
i
n
i
n
g
samp
l
e
s
T
e
st
i
n
g
S
a
mp
l
e
s
T
o
t
a
l
T
r
a
i
n
i
n
g
samp
l
e
s
T
e
st
i
n
g
S
a
mp
l
e
s
T
o
t
a
l
5
se
c
2
4
0
60
3
0
0
77
20
97
1
7
2
44
2
1
6
1
0
se
c
2
0
2
50
2
5
2
37
9
46
1
7
2
44
2
1
6
2
0
se
c
1
1
2
28
1
4
0
17
5
22
1
7
2
44
2
1
6
2
.
2
.
Aut
o
re
g
re
s
s
iv
e
m
o
del
li
ng
I
n
au
to
r
eg
r
es
s
i
v
e
m
o
d
el
t
h
e
p
r
esen
t
o
u
tp
u
t
o
f
a
n
y
t
i
m
e
s
er
ies
is
p
r
ed
icted
f
r
o
m
t
h
e
p
ast
o
u
tp
u
ts
.
B
asicall
y
,
it
f
it
s
an
o
p
ti
m
ized
cu
r
v
e
to
t
h
e
ex
i
s
ti
n
g
d
ata
p
o
in
ts
.
I
t
h
as
b
ee
n
u
s
ed
s
u
cc
e
s
s
f
u
l
l
y
i
n
v
ar
io
u
s
f
ield
s
s
u
c
h
as
s
p
ee
ch
p
r
o
ce
s
s
i
n
g
[
2
6
]
,
p
atter
n
r
ec
o
g
n
itio
n
[
2
7
]
an
d
b
io
m
ed
ical
s
i
g
n
al
p
r
o
ce
s
s
i
n
g
[
2
8
]
.
L
et’
s
a
s
s
u
m
e
a
ti
m
e
s
er
ies
Y
(
n
)
w
it
h
s
a
m
p
le
s
y
1
,
y
2
,
y
3
,
e
tc.
T
h
e
au
to
r
eg
r
ess
i
v
e
m
o
d
el
(
AR
(
p
)
)
,
h
av
i
n
g
o
r
d
er
p
,
is
d
ef
in
ed
as
Y
(
n
)
=
∑
(
)
(
−
)
+
(
)
=
1
(
1
)
Her
e,
th
e
p
is
t
h
e
o
r
d
er
o
f
th
e
m
o
d
el
,
ε
(
n
)
is
ze
r
o
m
ea
n
w
h
ite
n
o
is
e
s
eq
u
e
n
ce
w
it
h
a
v
ar
i
an
ce
o
f
σ
.
T
h
e
A
R
m
o
d
el
p
ar
a
m
eter
s
α
p
h
a
s
b
ee
n
ca
lcu
la
ted
u
s
i
n
g
B
u
r
g
’
s
m
e
th
o
d
,
b
ased
o
n
th
e
p
r
in
cip
les
o
f
m
i
n
i
m
izatio
n
o
f
f
o
r
w
ar
d
a
n
d
b
ac
k
w
ar
d
l
in
ea
r
p
r
ed
ictio
n
er
r
o
r
s
b
y
s
elec
tin
g
ap
p
r
o
p
r
iate
p
r
ed
ictio
n
co
ef
f
ic
ien
ts
s
u
b
j
ec
t
to
th
e
co
n
d
itio
n
t
h
at
t
h
e
y
m
u
s
t
s
ati
s
f
y
th
e
L
e
v
i
n
s
o
n
-
D
u
r
b
in
r
ec
u
r
s
iv
e
al
g
o
r
ith
m
.
T
h
ese
co
e
f
f
ic
i
en
ts
o
b
tai
n
ed
f
r
o
m
th
e
A
R
m
o
d
elli
n
g
o
f
t
h
e
E
C
G
s
ig
n
al
ar
e
u
s
ed
as f
ea
t
u
r
es to
t
h
e
ar
tif
ic
ial
n
e
u
r
al
n
et
w
o
r
k
.
2
.
3
.
Art
if
ici
a
l N
eura
l N
et
wo
rk
A
r
ti
f
icial
n
e
u
r
al
n
et
w
o
r
k
s
(
ANN)
ar
e
a
m
i
m
ic
o
f
t
h
e
b
io
l
o
g
ical
n
eu
r
o
n
s
p
r
ese
n
t
i
n
h
u
m
an
b
r
ain
.
T
h
ey
ar
e
u
s
ed
to
p
r
o
v
id
e
th
in
k
in
g
ca
p
ab
ilit
y
to
th
e
m
ac
h
i
n
es,
m
ak
i
n
g
th
e
m
s
m
ar
t
[
2
9
]
.
F
o
r
o
u
r
a
p
p
licatio
n
,
a
n
e
u
r
al
n
et
w
o
r
k
h
av
i
n
g
a
s
in
g
le
h
id
d
en
la
y
er
,
co
m
p
r
is
i
n
g
o
f
7
n
e
u
r
o
n
s
h
a
s
b
ee
n
u
s
ed
.
T
h
e
f
ea
tu
r
e
s
o
b
tain
ed
f
r
o
m
th
e
A
R
m
o
d
el
ar
e
u
s
ed
a
s
i
n
p
u
t
s
to
t
h
e
ANN.
T
h
e
o
u
tp
u
t
la
y
er
co
n
s
i
s
ti
n
g
o
f
3
n
e
u
r
o
n
s
h
a
s
b
ee
n
u
s
ed
to
s
u
cc
e
s
s
f
u
ll
y
cla
s
s
i
f
y
an
d
d
is
ti
n
g
u
i
s
h
b
et
w
ee
n
n
o
r
m
a
l Si
n
u
s
r
h
y
t
h
m
s
,
A
F a
n
d
A
F
L
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
7
,
No
.
2
,
J
u
n
e
20
1
8
:
90
–
94
92
Fig
u
r
e
1
.
Stru
ct
u
r
al
A
r
ti
f
ic
ial
Neu
r
al
N
et
w
o
r
k
Scaled
co
n
j
u
g
ate
g
r
ad
ien
t
b
ac
k
p
r
o
p
ag
atio
n
al
g
o
r
i
th
m
h
a
s
b
ee
n
u
s
ed
to
o
b
tain
t
h
e
co
r
r
ec
t
v
al
u
e
o
f
s
y
n
ap
ti
c
w
ei
g
h
ts
a
n
d
h
e
n
ce
m
i
n
i
m
ize
t
h
e
m
ea
n
s
q
u
ar
e
er
r
o
r
f
u
n
ctio
n
.
I
n
t
h
is
m
et
h
o
d
,
th
e
lear
n
in
g
r
ate
is
ad
j
u
s
ted
at
ea
ch
iter
atio
n
.
A
s
ea
r
c
h
i
s
m
ad
e
alo
n
g
t
h
e
co
n
j
u
g
ate
g
r
ad
ien
t
d
ir
ec
tio
n
to
d
eter
m
in
e
th
e
lear
n
i
n
g
r
ate,
w
h
ic
h
m
i
n
i
m
ize
s
th
e
p
er
f
o
r
m
a
n
ce
f
u
n
ctio
n
alo
n
g
t
h
at
li
n
e.
3.
RE
SU
L
T
S
A
ND
D
I
SCU
SS
I
O
N
An
ar
ti
f
icial
n
e
u
r
al
n
et
w
o
r
k
,
h
av
in
g
a
s
i
n
g
le
h
id
d
en
l
a
y
er
is
u
s
ed
to
d
if
f
er
en
t
iat
e
b
et
w
ee
n
f
ib
r
illatio
n
,
f
l
u
tter
a
n
d
n
o
r
m
al
s
in
u
s
E
C
G
s
i
g
n
al.
T
h
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
n
et
w
o
r
k
is
e
v
al
u
ated
b
y
ca
lcu
la
tin
g
p
ar
am
eter
s
li
k
e
s
p
ec
if
ic
it
y
,
s
e
n
s
it
iv
i
t
y
an
d
ac
c
u
r
ac
y
.
All
t
h
e
m
etr
ic
s
ar
e
b
ased
t
h
e
p
ar
a
m
eter
s
,
tr
u
e
p
o
s
iti
v
e
(
T
P
)
,
f
alse p
o
s
iti
v
e
(
FP
)
,
tr
u
e
n
eg
a
ti
v
e
(
T
N)
an
d
f
alse n
e
g
ati
v
e
(
FN)
.
Sen
s
iti
v
it
y
is
t
h
e
n
u
m
b
er
o
f
c
o
r
r
ec
tly
id
e
n
ti
f
ied
p
o
s
itiv
e
i
n
s
ta
n
ce
s
.
=
+
(
2
)
Sp
ec
if
icit
y
is
t
h
e
n
u
m
b
er
o
f
c
o
r
r
ec
tly
id
en
ti
f
ied
n
eg
a
tiv
e
i
n
s
tan
ce
s
.
=
+
(
3
)
A
cc
u
r
ac
y
is
t
h
e
co
r
r
ec
tn
e
s
s
o
f
th
e
s
y
s
te
m
,
i.e
.
t
h
e
clo
s
e
n
es
s
o
f
t
h
e
s
y
s
te
m
to
t
h
e
a
ctu
al
v
al
u
e.
T
h
ese
p
ar
am
eter
s
ar
e
ca
lcu
lated
f
r
o
m
th
e
d
ata
p
r
esen
ted
b
y
t
h
e
co
n
f
u
s
io
n
m
atr
ix
=
+
+
+
+
(
4
)
T
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
h
as
b
ee
n
d
ev
elo
p
ed
u
s
in
g
M
A
T
L
A
B
s
i
m
u
latio
n
e
n
v
ir
o
n
m
en
t.
T
h
e
s
y
s
te
m
is
tr
ai
n
ed
u
s
i
n
g
‘
n
n
s
tar
t’
GUI
,
b
y
u
s
in
g
a
d
ata
s
et
d
iv
id
ed
in
to
tr
ain
in
g
an
d
v
alid
atio
n
s
et
in
th
e
r
atio
o
f
4
:1
an
d
th
en
th
e
f
i
n
al
r
es
u
lt
s
ar
e
o
b
tai
n
ed
o
n
an
i
n
d
ep
en
d
en
t
test
s
et.
T
h
e
AR
m
o
d
el
is
tes
ted
b
y
v
ar
y
i
n
g
t
h
e
n
u
m
b
er
o
f
f
ea
t
u
r
es
to
o
b
tain
th
e
b
est
f
it.
T
ab
le
2
s
h
o
w
s
th
e
p
er
f
o
r
m
a
n
c
e
o
f
th
e
m
o
d
el
in
th
e
tr
ain
in
g
s
et
w
ith
v
aio
u
s
AR
f
ea
t
u
r
es (
p
)
,
f
o
r
a
5
s
ec
s
ig
n
al
s
a
m
p
le.
T
ab
le
2
.
P
er
f
o
r
m
a
n
ce
r
esu
l
t o
f
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
f
o
r
5
s
ec
ar
r
h
y
th
m
ia
s
s
a
m
p
le
p
S
e
n
si
t
i
v
i
t
y
S
p
e
c
i
f
i
c
i
t
y
A
c
c
u
r
a
c
y
(
%)
AF
N
S
R
A
F
L
AF
N
S
R
A
F
L
2
1
1
0
.
1
3
0
.
7
2
1
1
8
5
.
9
4
0
.
9
4
1
0
.
5
5
0
.
8
6
1
0
.
9
6
9
0
.
2
6
0
.
9
8
1
0
.
7
6
0
.
9
1
1
0
.
9
9
9
5
.
1
8
0
.
9
6
0
.
9
9
0
.
6
9
0
.
9
0
.
9
9
0
.
9
8
9
3
.
1
10
0
.
9
9
1
0
.
9
4
0
.
9
8
0
.
9
9
1
9
9
.
0
12
0
.
9
9
1
0
.
9
4
0
.
9
8
0
.
9
8
1
9
8
.
1
I
t
is
ev
id
e
n
t
f
r
o
m
T
ab
le
2
th
at
th
e
b
est
o
p
ti
m
u
m
p
er
f
o
r
m
an
c
e
is
o
b
tain
ed
f
o
r
n
u
m
b
er
o
f
f
e
atu
r
es
eq
u
a
l
to
1
0
.
T
h
e
p
er
f
o
r
m
a
n
ce
is
a
ls
o
ev
a
l
u
ated
f
o
r
d
if
f
er
e
n
t
ti
m
e
le
n
g
t
h
o
f
s
i
g
n
al
s
.
T
ab
le
3
co
m
p
ar
es
th
e
p
er
f
o
r
m
an
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
C
la
s
s
i
fica
tio
n
o
f A
tr
ia
l A
r
r
h
yth
mia
s
u
s
in
g
N
eu
r
a
l Netw
o
r
k
s
(
Ja
i U
tka
r
s
h
)
93
p
ar
am
eter
s
o
f
th
e
s
y
s
te
m
f
o
r
5
s
ec
,
1
0
s
ec
an
d
2
0
s
ec
s
a
m
p
les
o
n
tr
ain
i
n
g
s
et,
k
ee
p
i
n
g
t
h
e
n
u
m
b
er
o
f
f
ea
t
u
r
es a
s
1
0
.
T
ab
le
3
.
P
er
f
o
r
m
a
n
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
in
t
h
e
tr
ain
i
n
g
s
et
o
f
5
,
1
0
an
d
2
0
s
ec
d
u
r
atio
n
r
ec
o
r
d
in
g
s
(
k
ee
p
in
g
th
e
m
o
d
el
o
r
d
er
as 1
0
)
T
D
u
r
a
t
i
o
n
o
f
samp
l
e
(
se
c
)
S
e
n
si
t
i
v
i
t
y
S
p
e
c
i
f
i
c
i
t
y
A
c
c
u
r
a
c
y
(
%)
AF
N
S
R
A
F
L
AF
N
S
R
A
F
L
t
=
5
se
c
0
.
9
9
1
0
.
9
4
0
.
9
8
0
.
9
9
1
99
t
=
1
0
se
c
1
1
0
.
9
7
0
.
9
9
1
1
9
9
.
8
t
=
2
0
se
c
1
1
1
1
1
1
1
0
0
I
t
ca
n
b
e
s
ee
n
f
r
o
m
T
ab
le
3
th
at
tr
ai
n
i
n
g
s
e
t
ac
c
u
r
ac
y
is
b
etter
f
o
r
al
l
t
h
r
ee
ti
m
e
d
u
r
atio
n
s
a
m
p
les
.
T
h
e
ac
cu
r
ac
y
r
an
g
es
f
r
o
m
9
9
%
f
o
r
a
5
s
ec
tim
e
s
i
g
n
al
to
1
0
0
%
f
o
r
a
2
0
s
ec
ti
m
e
s
i
g
n
a
l.
T
h
e
p
er
f
o
r
m
a
n
ce
p
ar
am
eter
s
o
f
t
h
e
i
n
d
ep
en
d
en
t
test
s
et
f
o
r
5
s
ec
,
1
0
s
ec
a
n
d
2
0
s
ec
s
a
m
p
les
ar
e
p
r
esen
t
ed
in
T
ab
le
4
.
T
h
e
s
en
s
iti
v
i
t
y
r
ea
c
h
es
a
m
i
n
i
m
u
m
o
f
0
.
9
5
an
d
0
.
7
9
f
o
r
A
F
an
d
AF
L
r
esp
ec
tiv
el
y
f
o
r
th
e
5
s
ec
s
a
m
p
le.
T
h
e
s
p
ec
if
icit
y
r
ea
c
h
es
a
m
i
n
i
m
u
m
o
f
0
.
9
3
an
d
0
.
9
6
f
o
r
A
F
a
n
d
NS
R
r
esp
ec
ti
v
el
y
f
o
r
th
e
5
s
ec
s
a
m
p
le.
T
h
e
m
i
n
i
m
u
m
an
d
m
a
x
i
m
u
m
ac
c
u
r
ac
y
o
b
tain
ed
ar
e
9
4
.
3
%
an
d
1
0
0
%
r
esp
ec
tiv
el
y
.
T
h
e
p
er
f
o
r
m
an
ce
r
es
u
lt
s
o
f
th
e
test
s
et
s
f
o
r
t
h
e
th
r
ee
ca
s
es o
f
v
ar
y
i
n
g
ti
m
e
s
a
m
p
le
d
u
r
atio
n
ar
e
p
r
esen
ted
in
T
ab
le
s
5
,
6
,
a
n
d
7.
I
t is seen
f
r
o
m
T
ab
le
4
,
5
th
at
th
er
e
ar
e
s
o
m
e
m
i
s
clas
s
i
f
icatio
n
s
in
all
th
e
th
r
ee
ca
teg
o
r
ies
n
a
m
el
y
a
tr
ial
f
ib
r
illatio
n
,
atr
ia
l
f
l
u
tter
an
d
n
o
r
m
al
s
i
n
u
s
r
h
y
t
h
m
,
th
er
eb
y
l
i
m
itin
g
t
h
e
ac
cu
r
a
c
y
to
9
4
.
3
%.
T
ab
le
4
.
P
er
f
o
r
m
a
n
ce
r
esu
l
ts
o
n
ar
r
h
y
t
h
m
ia
s
in
d
ep
en
d
e
n
t te
s
t s
et
T
i
me
D
u
r
a
t
i
o
n
o
f
samp
l
e
(
se
c
)
S
e
n
si
t
i
v
i
t
y
S
p
e
c
i
f
i
c
i
t
y
A
c
c
u
r
a
c
y
(
%)
AF
N
S
R
A
F
L
AF
N
S
R
A
F
L
t
=
5
se
c
0
.
9
5
1
0
.
7
9
0
.
9
3
0
.
9
6
1
9
4
.
3
t
=
1
0
se
c
1
1
1
1
1
1
1
0
0
t
=
2
0
se
c
1
1
1
1
1
1
1
0
0
T
ab
le
5
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
t
h
e
i
n
d
ep
en
d
e
n
t
t
es
t
s
et
f
o
r
5
s
ec
a
r
r
h
y
th
m
ia
s
s
a
m
p
le
A
t
r
i
a
l
F
i
b
r
i
l
l
a
t
i
o
n
N
o
r
mal
S
i
n
u
s
R
h
y
t
h
m
A
t
r
i
a
l
F
l
u
t
t
e
r
A
t
r
i
a
l
F
i
b
r
i
l
l
a
t
i
o
n
57
0
4
N
o
r
mal
S
i
n
u
s
R
h
y
t
h
m
3
4
4
0
A
t
r
i
a
l
F
l
u
t
t
e
r
0
0
1
6
T
ab
le
6
.
I
n
d
ep
en
d
en
t te
s
t set
m
atr
i
x
f
o
r
1
0
s
ec
ar
r
h
y
th
m
ia
s
s
a
m
p
le
A
t
r
i
a
l
F
i
b
r
i
l
l
a
t
i
o
n
N
o
r
mal
S
i
n
u
s
R
h
y
t
h
m
A
t
r
i
a
l
F
l
u
t
t
e
r
A
t
r
i
a
l
F
i
b
r
i
l
l
a
t
i
o
n
50
0
0
N
o
r
mal
S
i
n
u
s
R
h
y
t
h
m
0
4
4
0
A
t
r
i
a
l
F
l
u
t
t
e
r
0
0
9
T
ab
le
7
.
I
n
d
ep
en
d
en
t te
s
t set
m
atr
i
x
f
o
r
2
0
s
ec
ar
r
h
y
th
m
ia
s
s
a
m
p
le
A
t
r
i
a
l
F
i
b
r
i
l
l
a
t
i
o
n
N
o
r
mal
S
i
n
u
s
R
h
y
t
h
m
A
t
r
i
a
l
F
l
u
t
t
e
r
A
t
r
i
a
l
F
i
b
r
i
l
l
a
t
i
o
n
28
0
0
N
o
r
mal
S
i
n
u
s
R
h
y
t
h
m
0
4
4
0
A
t
r
i
a
l
F
l
u
t
t
e
r
0
0
5
Fu
r
t
h
er
f
r
o
m
t
h
e
r
esu
lts
,
it
i
s
ev
id
en
t
th
at
,
t
h
er
e
ar
e
n
o
m
is
c
lass
if
ica
tio
n
s
in
ca
s
e
o
f
1
0
s
ec
an
d
2
0
s
ec
s
a
m
p
le
s
,
g
i
v
i
n
g
it
a
n
ac
cu
r
ac
y
o
f
1
0
0
%.
T
h
e
co
m
p
ar
is
o
n
r
esu
lt
s
o
f
d
i
f
f
er
en
t
d
u
r
atio
n
s
i
g
n
al
s
s
h
o
w
t
h
at
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
ef
f
icie
n
t i
n
clas
s
if
y
i
n
g
th
e
ar
r
y
t
h
i
m
ia
s
f
r
o
m
s
m
all
d
u
r
atio
n
s
i
g
n
al
s.
4.
CO
NCLU
SI
O
N
I
n
th
is
m
a
n
u
s
cr
ip
t,
an
ac
cu
r
ate
class
i
f
icat
io
n
tech
n
iq
u
e
h
as
b
ee
n
p
r
esen
ted
to
s
u
cc
es
s
f
u
ll
y
d
is
tin
g
u
is
h
b
et
w
ee
n
Fib
r
illati
o
n
,
Fl
u
tter
a
n
d
n
o
r
m
al
E
C
G
s
ig
n
al
s
.
Au
to
r
e
g
r
e
s
s
i
v
e
m
o
d
el
is
u
s
ed
to
e
x
tr
ac
t
r
elev
an
t
f
ea
t
u
r
es
f
r
o
m
t
h
e
E
C
G
s
i
g
n
als.
A
r
ti
f
ic
ial
n
e
u
r
al
n
et
w
o
r
k
h
as
b
ee
n
u
s
ed
to
tr
ain
t
h
e
m
o
d
el
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
m
o
d
el
h
as
b
ee
n
a
s
s
e
s
s
ed
i
n
d
i
f
f
er
en
t
l
en
g
t
h
o
f
s
ig
n
al
s
s
u
c
h
as
5
s
e
c,
1
0
s
ec
an
d
2
0
s
ec
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
7
,
No
.
2
,
J
u
n
e
20
1
8
:
90
–
94
94
T
h
e
tr
ain
in
g
s
et
an
d
test
s
et
ac
cu
r
ac
y
o
f
t
h
e
m
e
th
o
d
o
b
tain
ed
o
n
th
e
5
s
ec
ar
r
h
y
t
h
m
ias
s
a
m
p
le
is
9
9
%
an
d
9
4
.
3
% r
esp
ec
tiv
el
y
.
T
h
er
ef
o
r
e
,
th
e
s
y
s
te
m
ca
n
ea
s
il
y
b
e
u
s
e
d
to
p
r
ed
ict
atr
ial
ar
r
h
y
t
h
m
ias
w
it
h
a
n
E
C
G
s
i
g
n
a
l
o
f
s
m
all
d
u
r
atio
n
w
i
th
a
f
air
l
y
h
ig
h
ac
c
u
r
ac
y
.
RE
F
E
R
E
NC
E
S
[1
]
Ba
h
a
re
h
P
o
u
r
b
a
b
a
e
e
,
M
e
h
rsa
n
Ja
v
a
n
Ro
sh
tk
h
a
ri,
Kh
a
sh
a
y
a
rKh
o
ra
sa
n
i.
“
De
e
p
c
o
n
v
o
lu
ti
o
n
n
e
u
ra
l
n
e
tw
o
rk
s
a
n
d
lea
rn
in
g
ECG
fe
a
tu
re
s
f
o
r
sc
r
e
e
n
in
g
p
a
ro
x
y
s
m
a
l
A
tri
a
l
F
ib
ril
latio
n
P
a
ti
e
n
ts”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
y
ste
ms
,
M
a
n
a
n
d
Cy
b
e
rn
e
ti
c
s S
y
ste
ms
,
V
o
l
u
m
e
:
P
P
,
Iss
u
e
:
9
9
,
p
p
:
1
-
10,
2
0
1
7
[2
]
P
a
h
lm
O.
a
n
d
S
o
rn
m
o
L
.
,
“
S
o
f
t
w
a
r
e
QRS
d
e
te
c
ti
o
n
a
m
b
u
lato
ry
m
o
n
it
o
rin
g
-
A
R
e
v
ie
w”
.
M
e
d
.
Bi
o
l.
En
g
.
Co
mp
u
t.
V
o
l
2
2
,
p
2
8
9
-
2
9
7
,
1
9
8
4
.
[3
]
Ok
a
d
a
M
.
,
“
A
d
ig
it
a
l
f
il
ter
f
o
r
th
e
QRS
c
o
m
p
lex
d
e
tec
ti
o
n
”
,
IEE
E
tra
n
s.
Bi
o
me
d
.
En
g
r
.
,
V
o
l
.
BM
E
-
2
6
,
p
7
0
0
-
7
0
4
,
19
7
9
.
[4
]
V
a
n
Da
m
R
AA
F
,
Bre
k
e
l
m
a
n
s
F
EM
,
Du
isterh
o
u
t
JS,
“
A
h
ig
h
p
e
rf
o
r
m
a
n
c
e
M
icro
p
ro
c
e
ss
o
r
b
a
se
d
a
rrh
y
th
m
ias
m
o
n
it
o
r”
,
IEE
E
Co
mp
u
ter
s in
C
a
rd
.
S
o
c
.
,
p
4
4
9
-
4
5
2
,
1
9
8
1
.
[5
]
Re
d
d
y
BRS
,
M
u
rt
h
y
IS
N,
Ch
a
tt
e
rjee
P
C,
“
Rh
y
th
m
a
n
a
l
y
sis
u
sin
g
v
e
c
to
rc
a
rd
io
g
ra
m
”
.
IEE
E
T
ra
n
s.
Bi
o
me
d
.
E
n
g
r
.
,
V
o
l
BM
E
-
3
2
,
p
9
7
-
1
0
4
,
1
9
8
5
.
[6
]
T
re
m
b
la
y
G
.
Leb
lan
c
A
R.
“
Ne
a
r
-
o
p
ti
m
a
l
sig
n
a
l
p
re
p
ro
c
e
ss
in
g
f
o
r
p
o
siti
v
e
c
a
rd
iac
a
rrh
y
th
m
ias
id
e
n
t
if
ica
ti
o
n
”
,
IEE
E
T
ra
n
s.
B
io
me
d
.
En
g
r
.
,
Vo
l
BM
E
-
2
7
,
p
3
7
0
-
3
7
5
.
[7
]
Da
n
iel
Riv
e
ra
;
Cé
sa
r
V
e
ig
a
;
Ju
a
n
J.
Ro
d
ríg
u
e
z
-
A
n
d
in
a
;
Jo
sé
F
a
riñ
a
;
En
riq
u
e
G
a
rc
í
a
,
“
Us
in
g
su
p
p
o
rt
v
e
c
to
r
ma
c
h
in
e
fo
r
At
ria
l
fi
b
ri
ll
a
t
io
n
sc
re
e
n
in
g
,
”
2
0
1
7
IEE
E
2
6
th
In
tern
a
ti
o
n
a
l
S
y
m
p
o
siu
m
o
n
In
d
u
str
ial
El
e
c
tro
n
ic
s
(IS
IE)
,
P
a
g
e
s:
2
0
5
6
-
2
0
6
0
,
2
0
1
7
.
[8
]
G
.
Y.
L
ip
,
C.
M
.
Bre
c
h
i
n
,
a
n
d
D.
A
.
L
a
n
e
,
“
T
h
e
g
lo
b
a
l
b
u
r
d
e
n
o
f
a
tri
a
l
f
ib
ril
latio
n
a
n
d
stro
k
e
:
A
s
y
ste
m
a
ti
c
re
v
ie
w
o
f
th
e
e
p
i
d
e
m
io
lo
g
y
o
f
a
tri
a
l
f
ib
ril
latio
n
i
n
re
g
io
n
s
o
u
tsid
e
N
o
rth
Am
e
rica
a
n
d
Eu
r
o
p
e
,
”
CHE
S
T
J
.
,
v
o
l.
1
4
2
,
n
o
.
6
,
p
p
.
1
4
8
9
–
1
4
9
8
,
2
0
1
2
.
[9
]
M
.
C.
W
ij
ff
e
ls,
C.
J.
Kirc
h
h
o
f
,
R.
Do
rlan
d
,
a
n
d
M
.
A
.
A
ll
e
ss
ie,
“
A
tr
ial
f
ib
ril
latio
n
b
e
g
e
ts
a
tri
a
l
f
ib
ril
latio
n
.
A
stu
d
y
in
a
wa
k
e
c
h
ro
n
ica
ll
y
in
stru
m
e
n
ted
g
o
a
ts,”
Circ
u
la
t
io
n
,
v
o
l
.
9
2
,
p
p
.
1
9
5
4
–
1
9
6
8
,
Oc
t.
1
9
9
5
.
[1
0
]
M
.
S
tri
d
h
,
A
.
Bo
ll
m
a
n
n
,
S
.
B.
O
lsso
n
,
L
.
S
ö
rrn
m
o
,
“
De
tec
ti
o
n
a
n
d
f
e
a
tu
re
e
x
trac
ti
o
n
o
f
a
tri
a
l
tac
h
y
a
rrh
y
th
m
ias
,
”
IEE
E
E
n
g
g
.
i
n
M
e
d
icin
e
a
n
d
Bi
o
l
o
g
y
,
v
o
l
.
2
5
,
2
0
0
6
.
[1
1
]
S
tri
d
t
h
,
L
.
S
o
rn
m
o
,
“
S
h
a
p
e
Ch
a
ra
c
teriz
a
ti
o
n
o
f
a
tri
a
l
f
ib
ril
latio
n
u
sin
g
ti
m
e
-
f
r
e
q
u
e
n
c
y
a
n
a
l
y
sis
,
”
Co
mp
u
ter
s
in
Ca
rd
io
lo
g
y
,
v
o
l.
2
9
,
p
p
.
1
7
-
2
0
,
2
0
0
2
.
[1
2
]
K.T
a
ten
o
,
L
.
G
las
s,
“
A
u
to
m
a
ti
c
d
e
tec
ti
o
n
o
f
a
tri
a
l
f
ib
ril
latio
n
u
sin
g
th
e
c
o
e
f
f
icie
n
t
o
f
v
a
ria
ti
o
n
a
n
d
d
e
n
sity
h
isto
g
ra
m
s o
f
RR an
d
Δ
RR i
n
terv
a
ls
,
”
M
e
d
.
Bi
o
l.
E
n
g
g
.
C
o
mp
u
ti
n
g
,
v
o
l.
3
9
,
p
p
.
6
6
4
-
6
7
1
,
2
0
0
1
.
[1
3
]
B.
L
o
g
a
n
,
J.
He
a
le
y
,
“
Ro
b
u
st
d
e
tec
ti
o
n
o
f
a
tri
a
l
f
ib
ril
latio
n
f
o
r
a
lo
n
g
term
tele
m
o
n
it
o
rin
g
sy
ste
m
,
”
Co
mp
u
ter
s
in
Ca
rd
io
lo
g
y
,
p
p
6
1
9
-
6
2
,
2
0
0
5
.
[1
4
]
B.
Yo
u
n
g
,
D.
Bro
d
n
ick
,
R.
S
p
a
u
ld
i
n
g
,
“
A
C
o
mp
a
ra
ti
v
e
S
tu
d
y
o
f
a
Hid
d
e
n
M
a
rk
o
v
M
o
d
e
l
De
t
e
c
to
r
fo
r
At
ri
a
l
Fi
b
rill
a
ti
o
n
,
”
P
r
o
c
e
e
d
in
g
s o
f
th
e
1
9
9
9
IE
EE
S
ig
n
a
l
P
ro
c
e
ss
in
g
S
o
c
iety
W
o
rk
sh
o
p
,
p
p
4
6
8
-
4
7
6
,
1
9
9
9
.
[1
5
]
R.
M
a
b
r
o
u
k
i,
B.
Kh
a
d
d
o
u
m
i,
M
.
S
a
y
a
d
i,
“
No
n
li
n
e
a
r
S
ta
ti
st
ica
l
M
e
th
o
d
s
f
o
r
At
ria
l
Fi
b
rill
a
ti
o
n
De
tec
ti
o
n
o
n
e
lec
tro
c
a
rd
io
g
r
a
m
,
”
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
El
e
c
tri
c
a
l
S
c
ien
c
e
s a
n
d
T
e
c
h
n
o
l
o
g
ies
,
M
a
g
h
re
b
,
2
0
1
4
.
[1
6
]
Do
k
u
r,
Z.
,
Ölm
e
z
,
T
.
,
Ya
z
g
a
n
,
E.
,
“
Co
m
p
a
riso
n
o
f
d
isc
re
te
w
a
v
e
let
a
n
d
F
o
u
rier
T
ra
n
sf
o
r
m
s
f
o
r
EC
G
b
e
a
t
Clas
sif
ic
a
ti
o
n
”
,
IEE
E
El
e
c
tro
n
ics
L
e
tt
e
rs
On
li
n
e
N
o
:
1
9
9
9
1
0
9
5
,
1
9
9
9
.
[1
7
]
S
.
Isa
a
c
Niwa
s,
R.
S
h
a
n
t
h
a
S
e
lv
a
Ku
m
a
ri,
V
.
S
a
d
a
siv
a
m
,
“
Arti
fi
c
ia
l
Ne
u
ra
l
Ne
two
rk
Ba
se
d
Au
t
o
ma
ti
c
C
a
rd
i
a
c
Ab
n
o
rm
a
li
ti
e
s Cl
a
ss
if
ica
ti
o
n
”
,
IE
EE
,
ICCIM
A
’0
5
,
2
0
0
5
.
[1
8
]
B.
G
.
Ce
ll
e
r,
“
L
o
w
Co
mp
u
t
a
ti
o
n
a
l
Co
st
Cl
a
ss
if
ier
s
fo
r
ECG
Di
a
g
n
o
sis
Us
in
g
Ne
u
ra
l
Ne
tw
o
rk
s
”
,
P
r
o
c
e
e
d
in
g
s
o
f
th
e
20
th
A
n
n
u
a
l
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
r
e
n
c
e
o
f
th
e
IEE
E
En
g
i
n
e
e
rin
g
in
M
e
d
icin
e
a
n
d
Bi
o
lo
g
y
S
o
c
iet
y
,
V
o
l
.
2
0
,
B
o
3
,
1
9
9
8
.
[1
9
]
Z.
Do
k
u
r,
T
.
Ölm
e
z
,
“
EC
G
B
e
a
t
Clas
si
f
ica
ti
o
n
b
y
a
No
v
e
l
H
y
b
rid
Ne
u
ra
l
Ne
t
w
o
rk
”
,
Co
m
p
u
ter
M
e
th
o
d
s
a
n
d
P
r
o
g
ra
m
s in
Bio
m
e
d
icin
e
,
El
se
v
ie
r
,
2
0
0
0
.
[2
0
]
Do
k
u
r,
Z.
,
Ölm
e
z
,
T
.
,
Ya
z
g
a
n
,
E,
“
Cla
ss
if
ica
ti
o
n
o
f
ECGW
a
v
e
fo
rm
s
Us
in
g
a
No
v
e
l
Ne
u
ra
l
Ne
two
r
k
”
,
2
0
th
A
n
n
u
a
l
In
t.
C
o
n
f
e
re
n
c
e
o
f
th
e
IEE
E
-
EM
BS
,
v
o
l.
2
0
,
n
o
3
,
p
p
.
1
6
1
6
-
1
6
1
9
,
Ho
n
g
Ko
n
g
,
1
9
9
8
.
[2
1
]
Ka
n
g
-
P
i
n
g
L
in
;
W
.
H.
Ch
a
n
g
,
“
A
tec
h
n
iq
u
e
fo
r
a
u
to
m
a
ted
a
rr
y
th
mi
a
d
e
tec
ti
o
n
o
f
Ho
l
ter
ECG
”
.
P
r
o
c
e
e
d
in
g
s
o
f
1
7
t
h
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
f
th
e
En
g
in
e
e
rin
g
in
M
e
d
icin
e
a
n
d
Bi
o
l
o
g
y
S
o
c
iet
y
,
V
o
l
u
m
e
:
1
,
P
a
g
e
s:
1
8
3
-
1
8
4
,
1
9
9
5
.
[2
2
]
S
e
ç
il
Ze
y
b
e
k
o
g
lu
,
M
e
h
m
e
d
Öz
k
a
n
,
“
Cla
ss
if
ica
ti
o
n
o
f
ECG
Arrh
y
th
mia
s
b
e
a
ts
u
sin
g
Arti
fi
c
ia
l
Ne
u
ra
l
Ne
two
rk
s
”
,
2
0
1
0
1
5
t
h
Na
ti
o
n
a
l
Bi
o
m
e
d
ica
l
En
g
in
e
e
rin
g
M
e
e
ti
n
g
,
P
a
g
e
s:
1
-
4,
2
0
1
0
.
[2
3
]
T
h
e
M
IT
-
BIH
A
tri
a
l
F
ib
ril
lati
o
n
d
a
tab
a
se
[o
n
li
n
e
]
.
A
v
a
il
a
b
le:
h
tt
p
:
//
p
h
y
sio
n
e
t
.
o
rg
/
p
h
y
sio
b
a
n
k
/d
a
tab
a
se
/
m
it
d
b
/
[2
4
]
T
h
e
M
IT
-
BIH No
r
m
a
l
S
in
u
s Rh
y
th
m
d
a
tab
a
se
[
o
n
li
n
e
]
.
A
v
a
il
a
b
le:
h
tt
p
:
//
p
h
y
sio
n
e
t
.
o
rg
/p
h
y
sio
b
a
n
k
/d
a
tab
a
se
/
m
it
d
b
/
[2
5
]
T
h
e
M
IT
-
BIH arrh
y
th
m
ia d
a
tab
a
se
[
o
n
li
n
e
]
.
A
v
a
il
a
b
le:
h
tt
p
:
//
p
h
y
si
o
n
e
t.
o
rg
/p
h
y
sio
b
a
n
k
/d
a
tab
a
se
/m
it
d
b
/
[2
6
]
S
a
m
m
ie
G
il
e
s
;
Jill
Ba
r
f
ield
,
“
Au
to
re
g
re
ss
ive
M
o
d
e
li
n
g
o
f
L
a
y
e
re
d
M
u
lt
i
-
M
e
d
iu
m
fo
r
EM
S
ig
n
a
lP
r
o
c
e
ss
in
g
”
.
2
0
0
7
T
h
irt
y
-
Nin
th
S
o
u
t
h
e
a
ste
rn
S
y
m
p
o
siu
m
o
n
S
y
ste
m
T
h
e
o
r
y
,
P
a
g
e
s:
68
-
7
0
,
2
0
0
7
.
[2
7
]
L
izh
i
P
a
n
;
Di
n
g
g
u
o
Zh
a
n
g
;
X
i
n
ju
n
S
h
e
n
g
;
X
ian
g
y
a
n
g
Zh
u
,
“
Res
id
u
a
ls o
f
a
u
to
re
g
re
ss
ive
mo
d
e
l
p
ro
v
id
in
g
a
d
d
it
i
o
n
a
l
in
fo
rm
a
ti
o
n
f
o
r fea
tu
re
e
x
tra
c
ti
o
n
o
f
p
a
tt
e
rn
re
c
o
g
n
it
i
o
n
-
b
a
se
d
my
o
e
lec
tric c
o
n
tro
l
”
.
2
0
1
5
3
7
th
A
n
n
u
a
l
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
f
th
e
IEE
E
En
g
i
n
e
e
rin
g
in
M
e
d
icin
e
a
n
d
Bi
o
l
o
g
y
S
o
c
i
e
t
y
(EM
BC)
,
P
a
g
e
s:
7
2
7
0
-
7
2
7
3
,
2
0
1
5
.
[2
8
]
M
.
J.
Ca
ss
id
y
;
W
.
D.
P
e
n
n
y
,
“
Ba
y
e
sia
n
n
o
n
sta
ti
o
n
a
ry
a
u
to
re
g
re
ss
iv
e
m
o
d
e
ls
f
o
r
b
i
o
m
e
d
ica
l
sig
n
a
l
a
n
a
ly
sis
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Bi
o
me
d
ica
l
En
g
i
n
e
e
rin
g
,
Vo
lu
m
e
:
4
9
,
Iss
u
e
:
10
,
P
a
g
e
s:
1142
-
1
1
5
2
,
2
0
0
2
.
[2
9
]
“
Ne
u
ra
l
n
e
tw
o
rk
De
si
g
n
”
-
M
.
T
.
Ha
g
a
n
,
B.
De
m
u
th
&
M
.
Be
a
le,
Th
o
m
so
n
L
e
a
rn
in
g
,
2
0
0
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.