I
A
E
S
I
n
t
e
r
n
at
io
n
al
Jou
r
n
al
of
A
r
t
if
ic
ia
l
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
V
ol
.
11
, N
o.
1
,
M
a
r
c
h
2022
, pp.
345
~
355
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
11
.i
1
.pp
345
-
355
345
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
E
f
f
e
c
t
i
v
e
p
r
e
d
i
c
t
i
ve
m
od
e
l
l
i
n
g f
or
c
or
on
ar
y ar
t
e
r
y d
i
se
ase
s
u
si
n
g s
u
p
p
or
t
ve
c
t
or
m
ac
h
i
n
e
K
u
n
c
ah
yo S
e
t
yo N
u
gr
oh
o, A
n
an
t
h
a Y
u
ll
ia
n
S
u
k
m
ad
e
w
a, A
n
gga Vid
ia
n
t
o, Wayan
F
ir
d
au
s
M
ah
m
u
d
y
D
e
pa
r
t
m
e
nt
of
I
nf
or
m
a
t
i
c
s
E
ngi
ne
e
r
i
ng,
F
a
c
ul
t
y of
C
om
put
e
r
S
c
i
e
nc
e
, B
r
a
w
i
j
a
ya
U
ni
ve
r
s
i
t
y, M
a
l
a
ng, I
ndone
s
i
a
A
r
t
ic
le
I
n
f
o
A
B
S
T
R
A
C
T
A
r
ti
c
le
h
is
to
r
y
:
R
e
c
e
iv
e
d
J
un 21, 2021
R
e
vi
s
e
d
D
e
c
24, 2021
A
c
c
e
pt
e
d
J
a
n 2, 2022
Coronar
y
artery
disease
(CAD)
is
a
catego
ry
of
cardiov
ascula
r
disea
se
that
causes
the
highest
mortality
rate
in
the
world.
CAD
occurs
due
to
plaque
build
-
up
on
the
walls
of
the
arteries
that
supply
blood
to
the
heart
an
d
other
organs
of
the
body.
To
contro
l
the
mortality
rate,
a
practical
model
that
is
capable
of
predicti
ng
CAD
is
needed.
Machine
learning
approache
s
have
been
used
in
solving
various
problems
in
various
domains,
inc
luding
biomedicine.
However,
real
-
world
data
often
has
an
unbalanced
class
di
stribution
that
can
interfere
with
classifier
performanc
e.
In
additio
n,
data
has
many
features
to
process.
This
study
focuses
on
effective
mo
deling
capable
of
predicti
ng
CAD
using
feature
selectio
n
to
handle
high
dimensional
data
and
feature
resampling
to
handle
unbalanced
data.
Feature
selection is very e
ffective by e
liminating irrelevant f
eatures
from the tr
aining
data.
Hyperparameter
tuning
is
also
done
to
find
the
best
combinat
ion
of
parameters
in
support
vector
machines
(
SVM
)
.
Our
results
show
t
hat
the
SVM
cross
-
validated
ten
times
has
a
more
accurate
training
result.
Further
more,
the
grid
searc
h
on
SVM
cross
-
validated
ten
times
had
more
accurate trai
ning m
odel resu
lts and
achieved 8
8% accuracy on
the tes
t data.
K
e
y
w
o
r
d
s
:
C
or
ona
r
y a
r
te
r
y di
s
e
a
s
e
pr
e
di
c
ti
on
E
f
f
e
c
ti
ve
s
uppor
t
ve
c
to
r
m
a
c
hi
ne
s
G
ood
s
uppor
t
ve
c
to
r
m
a
c
hi
ne
s
pa
r
a
m
e
te
r
s
This is an
open
acce
ss artic
le unde
r the
CC BY
-
SA
license.
C
or
r
e
s
pon
di
n
g A
u
th
or
:
K
unc
a
hyo S
e
ty
o
N
ugr
oho
D
e
pa
r
tm
e
nt
of
I
nf
or
m
a
ti
c
s
E
ngi
ne
e
r
in
g, F
a
c
ul
ty
of
C
om
put
e
r
S
c
ie
nc
e
, B
r
a
w
ij
a
ya
U
ni
ve
r
s
it
y
J
a
la
n V
e
te
r
a
n
,
N
o. 8, L
ow
okw
a
r
u, M
a
la
ng, E
a
s
t
J
a
va
65145, I
n
done
s
ia
E
m
a
il
:
ks
nugr
oho26@
gm
a
il
.c
om
1.
I
N
T
R
O
D
U
C
T
I
O
N
T
he
he
a
r
t
is
a
n
e
s
s
e
nt
ia
l
or
ga
n
in
th
e
c
a
r
di
ova
s
c
ul
a
r
s
ys
te
m
th
a
t
r
e
qui
r
e
s
a
s
uppl
y
of
bl
ood
th
a
t
c
ont
a
in
s
oxyge
n.
T
h
e
c
or
ona
r
y
c
ir
c
ul
a
ti
on
s
uppl
ie
s
bl
ood
to
th
e
he
a
r
t.
T
he
a
or
ta
di
vi
de
s
in
to
two
m
a
jo
r
c
or
ona
r
y
a
r
te
r
ie
s
,
e
a
c
h
of
w
hi
c
h
br
a
nc
he
s
of
f
in
to
s
m
a
ll
e
r
a
r
te
r
ie
s
a
nd
s
uppl
i
e
s
bl
ood
to
th
e
e
nt
ir
e
he
a
r
t
m
us
c
le
[
1]
.
C
a
r
di
ova
s
c
ul
a
r
di
s
e
a
s
e
(
C
V
D
)
is
a
gr
oup
of
di
s
e
a
s
e
s
.
C
or
ona
r
y
he
a
r
t
di
s
e
a
s
e
(
C
H
D
)
,
c
or
ona
r
y
a
r
te
r
y
di
s
e
a
s
e
(
C
A
D
)
,
a
nd
a
c
ut
e
c
or
ona
r
y
s
yndr
om
e
(
A
C
S
)
a
r
e
in
c
lu
de
d
in
C
V
D
[
2]
.
C
A
D
ha
ppe
ns
b
e
c
a
u
s
e
of
pl
a
que
bui
ld
-
up
in
th
e
w
a
ll
of
a
r
te
r
ie
s
th
a
t
s
uppl
y
bl
ood
to
th
e
he
a
r
t
a
nd
ot
he
r
pa
r
ts
of
th
e
[
3
]
.
T
he
c
ondi
ti
on
known
a
s
a
th
e
r
os
c
le
r
os
is
oc
c
ur
s
w
he
n
pl
a
que
be
gi
ns
to
a
c
c
um
ul
a
te
in
th
e
s
e
a
r
te
r
ie
s
.
A
s
th
e
pl
a
que
ha
r
de
ns
,
th
e
c
or
ona
r
y
a
r
te
r
ie
s
na
r
r
ow
,
de
c
r
e
a
s
in
g
th
e
bl
ood
s
u
ppl
y
to
th
e
he
a
r
t.
A
bl
ood
c
lo
t
on
th
e
pl
a
que
'
s
s
ur
f
a
c
e
m
a
y
oc
c
ur
if
it
r
upt
ur
e
s
.
I
n
th
e
m
a
jo
r
it
y
of
s
it
ua
ti
ons
,
a
bi
g
bl
ood
c
lo
t
c
a
n
to
ta
ll
y
s
to
p
th
e
c
or
ona
r
y
a
r
te
r
ie
s
'
bl
ood
f
lo
w
.
A
he
a
r
t
a
tt
a
c
k,
if
le
f
t
unt
r
e
a
te
d,
c
a
n
r
e
s
ul
t
in
m
a
jo
r
he
a
lt
h
c
ons
e
que
nc
e
s
a
nd,
in
th
e
w
or
s
t
-
c
a
s
e
s
c
e
na
r
io
, de
a
th
. A
s
a
r
e
s
ul
t,
c
a
r
di
ova
s
c
ul
a
r
di
s
e
a
s
e
i
s
th
e
l
e
a
di
ng c
a
u
s
e
of
de
a
th
gl
oba
ll
y
[
4]
.
T
he
r
e
is
a
n
a
de
qu
a
te
ne
e
d
f
or
th
e
e
a
r
ly
de
te
c
ti
on
of
pa
ti
e
nt
s
w
it
h
C
A
D
.
A
m
a
c
hi
ne
le
a
r
ni
ng
a
ppr
oa
c
h
c
a
n
s
ol
ve
pr
obl
e
m
s
in
th
e
bi
om
e
di
c
a
l
dom
a
in
[
5]
–
[
8]
.
M
a
c
hi
ne
le
a
r
ni
ng
gi
ve
s
th
e
c
om
put
e
r
a
bi
li
ty
to
le
a
r
n
a
nd
im
pr
ove
f
r
om
e
xpe
r
ie
nc
e
a
ut
om
a
ti
c
a
ll
y.
M
a
c
h
in
e
le
a
r
ni
ng
a
lg
or
it
hm
s
ha
ve
s
e
ve
r
a
l
m
a
jo
r
c
a
te
gor
ie
s
b
a
s
e
d
on
th
e
ir
le
a
r
ni
ng
a
ppr
oa
c
h,
in
put
a
nd
o
ut
put
da
ta
,
a
nd
pr
obl
e
m
ty
pe
:
s
upe
r
vi
s
e
d,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
11
, N
o.
1
,
M
a
r
c
h
20
22
:
345
-
355
346
uns
upe
r
vi
s
e
d,
a
nd
r
e
in
f
or
c
e
m
e
nt
le
a
r
ni
ng
[
9]
.
S
upp
or
t
ve
c
to
r
m
a
c
hi
ne
s
(
S
V
M
s
)
us
e
d
in
s
upe
r
vi
s
e
d
le
a
r
ni
ng
ha
ve
be
e
n s
how
n t
o be
e
xt
r
e
m
e
ly
e
f
f
e
c
ti
ve
a
t
s
ol
vi
ng c
la
s
s
if
ic
a
ti
on pr
obl
e
m
s
i
n
a
va
r
ie
ty
o
f
bi
om
e
di
c
a
l
f
ie
ld
s
[
6]
, [
10]
–
[
12]
.
T
he
r
e
a
r
e
m
a
ny
s
tu
di
e
s
c
onduc
te
d
to
di
a
gnos
e
C
A
D
w
it
h
m
a
c
hi
ne
le
a
r
ni
ng
in
r
e
c
e
nt
ye
a
r
s
.
T
he
m
os
t
w
id
e
ly
us
e
d
da
t
a
s
e
t
in
C
A
D
di
a
gnos
i
s
is
th
e
Z
-
A
li
z
a
d
e
h
s
a
ni
da
ta
s
e
t
[
13]
.
U
s
in
g
th
is
da
ta
s
e
t,
[
14]
a
ppl
ie
d
da
ta
m
in
in
g
te
c
hni
que
s
to
di
a
gnos
e
C
A
D
ba
s
e
d
on
th
e
s
ym
pt
om
s
a
nd
c
ha
r
a
c
te
r
is
ti
c
s
of
th
e
pa
ti
e
nt
'
s
E
C
G
.
T
he
ir
r
e
s
e
a
r
c
h
us
e
d
s
e
que
nt
ia
l
m
in
im
a
l
opt
im
iz
a
ti
on
(
S
M
O
)
a
nd
na
ïv
e
ba
ye
s
(
N
B
)
c
la
s
s
if
ie
r
a
nd
a
c
om
bi
na
ti
on
of
bot
h
to
di
a
gnos
e
C
A
D
.
T
e
s
ti
ng
w
it
h
10
-
f
ol
d
c
r
os
s
-
va
li
da
ti
on
s
how
s
th
a
t
th
e
c
om
bi
na
ti
on
of
S
M
O
-
na
ïv
e
ba
ye
s
i
s
s
upe
r
io
r
by a
c
hi
e
vi
ng mor
e
t
ha
n 88.52%
a
c
c
ur
a
c
y t
ha
n S
M
O
of
86.95%
a
nd na
ïv
e
ba
ye
s
of
87.22%
.
I
n
a
not
he
r
s
tu
dy,
[
15]
us
in
g
S
M
O
,
na
ïv
e
ba
ye
s
,
ba
ggi
ng
w
it
h
S
M
O
,
a
nd
ne
ur
a
l
ne
twor
k
to
di
a
gnos
e
th
e
s
a
m
e
di
s
e
a
s
e
.
I
nf
or
m
a
ti
on
ga
in
is
us
e
d
to
de
te
r
m
in
e
w
hi
c
h
f
e
a
tu
r
e
s
a
r
e
m
os
t
e
f
f
e
c
ti
ve
f
or
di
a
gnos
in
g
C
A
D
.
A
s
a
r
e
s
ul
t,
S
M
O
w
it
h
in
f
or
m
a
ti
on
ga
in
obt
a
in
e
d
th
e
be
s
t
p
e
r
f
or
m
a
nc
e
w
it
h
a
n
a
c
c
ur
a
c
y
of
94.08%
.
A
li
z
a
de
h
s
a
ni
e
t
al
.
us
e
d
th
e
f
e
a
tu
r
e
s
e
le
c
ti
on
te
c
hni
que
us
e
d
in
N
B
,
C
4.5,
a
nd
S
V
M
to
di
a
gnos
e
C
A
D
.
U
s
in
g
10
-
f
ol
d
c
r
os
s
-
va
li
da
ti
on,
S
V
M
ha
s
th
e
hi
ghe
s
t
a
c
c
ur
a
c
y
of
96.40%
[
13]
.
T
o
in
c
r
e
a
s
e
a
c
c
ur
a
c
y,
[
16]
us
e
d
r
a
ndom
tr
e
e
s
(
R
T
)
,
de
c
is
io
n
tr
e
e
(
D
T
)
,
S
V
M
,
a
n
d
c
hi
-
s
qua
r
e
d
a
ut
om
a
ti
c
in
te
r
a
c
ti
on
de
te
c
ti
on
(
C
H
A
I
D
)
to
s
e
le
c
t
f
e
a
tu
r
e
s
ba
s
e
d
on
pr
e
de
f
in
e
d
c
r
it
e
r
ia
f
o
r
C
A
D
di
a
gnos
is
.
R
a
ndom
tr
e
e
s
a
r
e
th
e
be
s
t
m
e
th
od
by
s
e
le
c
ti
ng
40
s
ig
ni
f
ic
a
nt
f
e
a
tu
r
e
s
a
nd
br
in
gi
ng
out
a
n
a
c
c
ur
a
c
y
of
91.47%
[
17]
us
in
g
hybr
id
P
C
A
,
D
T
,
a
nd
f
ir
e
f
ly
opt
im
iz
a
ti
on
te
c
hni
que
s
to
opt
im
iz
e
th
e
a
c
c
ur
a
c
y
of
e
xi
s
ti
ng
m
ode
ls
.
T
he
P
C
A
a
lg
or
it
hm
is
us
e
d
to
e
xt
r
a
c
t
f
e
a
tu
r
e
s
,
th
e
f
ir
e
f
ly
op
ti
m
iz
a
ti
on
te
c
hni
que
is
u
s
e
d
to
opt
im
iz
e
th
e
f
e
a
tu
r
e
s
e
le
c
ti
on, a
nd
D
T
is
us
e
d
to
c
la
s
s
if
y
th
e
da
ta
.
T
he
y
a
c
hi
e
ve
93%
a
c
c
ur
a
c
y
w
it
h
a
lo
w
c
la
s
s
if
ic
a
ti
on
e
r
r
or
r
a
te
a
ls
o
lo
w
f
a
ls
e
pos
it
iv
e
a
nd ne
ga
ti
ve
r
a
te
s
.
O
th
e
r
s
tu
di
e
s
ha
ve
a
l
s
o
be
e
n
c
onduc
t
e
d
on
th
e
p
r
e
di
c
ti
on
of
di
s
e
a
s
e
in
th
e
bi
om
e
di
c
a
l
f
ie
ld
.
S
V
M
is
us
e
d
to
pr
e
di
c
t
di
a
be
te
s
a
nd
pr
e
-
di
a
be
te
s
[
10]
.
T
h
e
S
V
M
m
od
e
l
is
us
e
d
to
id
e
nt
if
y
c
ha
r
a
c
t
e
r
is
ti
c
s
th
a
t
be
s
t
c
la
s
s
if
y
in
di
vi
dua
ls
in
to
di
f
f
e
r
e
nt
di
a
be
te
s
s
ubt
ype
s
.
T
he
ir
m
ode
l
got
83.47%
f
or
de
te
c
ti
on
of
di
a
gnos
e
d
di
a
be
te
s
or
di
a
gnos
e
d
di
a
be
te
s
c
om
pa
r
e
d
to
[
18]
m
ode
l
th
a
t
go
t
82.1%
.
I
n
th
is
r
e
s
e
a
r
c
h,
th
e
y
c
onc
lu
de
S
V
M
is
a
pr
om
is
in
g m
ode
l
f
or
de
te
c
ti
ng a
c
om
pl
e
x di
s
e
a
s
e
us
in
g c
o
m
m
on a
nd s
im
pl
e
va
r
ia
bl
e
s
. A
c
c
or
di
ng t
o [
11]
,
[
19]
S
V
M
ha
s
s
u
pe
r
io
r
a
c
c
ur
a
c
y
w
he
n
pr
e
di
c
ti
ng
he
a
r
t
di
s
e
a
s
e
,
di
a
be
te
s
,
a
nd
pa
r
ki
ns
on’
s
di
s
e
a
s
e
.
S
V
M
w
a
s
a
ls
o r
e
por
te
d t
o obta
in
be
tt
e
r
a
c
c
ur
a
c
y t
ha
n r
a
ndom f
or
e
s
t
in
br
e
a
s
t
c
a
n
c
e
r
pr
e
di
c
ti
on
[
20]
.
M
a
c
hi
ne
le
a
r
ni
ng
a
s
s
um
in
g
a
ba
la
n
c
e
d
num
be
r
of
in
s
ta
nc
e
s
in
e
a
c
h
c
la
s
s
.
W
h
e
n
us
in
g
unba
la
nc
e
d
da
ta
c
a
n
le
a
d
to
in
a
c
c
ur
a
te
m
ode
l
pr
e
di
c
ti
on
r
e
s
ul
ts
.
S
ynt
he
ti
c
m
in
or
it
y
ove
r
s
a
m
pl
in
g
te
c
hni
que
(
S
M
O
T
E
)
a
nd
a
da
pt
iv
e
s
ynt
he
ti
c
(
A
D
A
S
Y
N
)
s
a
m
pl
in
g
a
r
e
a
lt
e
r
na
ti
ve
s
to
ove
r
c
om
e
unba
la
nc
e
d
d
a
ta
by
c
r
e
a
ti
ng
s
ynt
he
ti
c
da
ta
in
t
he
m
in
or
it
y c
la
s
s
[
21]
. S
M
O
T
E
, w
hi
c
h i
s
i
nt
e
gr
a
te
d w
it
h t
he
pr
e
di
c
ti
on mode
l
is
r
e
por
te
d t
o
im
pr
ove
th
e
pr
e
di
c
ti
on
m
ode
l'
s
pe
r
f
or
m
a
nc
e
[
22]
,
[
23
]
.
I
n
C
A
D
pr
e
di
c
ti
on,
S
M
O
T
E
on
a
r
ti
f
ic
ia
l
ne
ur
a
l
ne
twor
ks
,
D
T
,
a
nd
S
V
M
s
how
e
d
a
n
in
c
r
e
a
s
e
in
th
e
a
c
c
u
r
a
c
y
ob
ta
in
e
d
f
r
om
th
e
or
ig
in
a
l
da
ta
[
24]
.
M
e
a
n
w
hi
le
,
[
25]
us
in
g
A
D
A
S
Y
N
w
it
h
S
V
M
to
di
a
gnos
e
P
a
r
ki
ns
on'
s
di
s
e
a
s
e
e
f
f
e
c
ti
ve
ly
.
B
ot
h
s
tu
di
e
s
do
not
e
m
pl
oy f
e
a
tu
r
e
s
e
le
c
ti
on t
o de
te
r
m
in
e
t
he
m
os
t
e
s
s
e
nt
i
a
l
f
e
a
tu
r
e
s
f
or
out
put
pr
e
di
c
ti
on.
B
a
s
e
d
on
pr
e
vi
ous
s
tu
di
e
s
,
a
c
om
bi
na
ti
on
of
f
e
a
tu
r
e
s
e
le
c
ti
on
a
nd
f
e
a
t
ur
e
r
e
s
a
m
pl
in
g
in
C
A
D
pr
e
di
c
ti
on
ha
s
ne
ve
r
be
e
n
done
be
f
or
e
.
B
ot
h
te
c
hni
que
s
a
r
e
r
e
por
te
d
to
im
p
r
ove
th
e
pe
r
f
or
m
a
nc
e
of
th
e
r
e
s
ul
ti
ng
m
ode
l.
T
h
e
m
a
in
c
ont
r
ib
ut
io
n
of
th
is
s
tu
dy
i
s
th
a
t
w
e
pr
opos
e
a
f
r
a
m
e
w
or
k
f
or
bui
ld
in
g
a
n e
f
f
e
c
ti
ve
m
ode
l
us
in
g f
e
a
tu
r
e
s
e
le
c
ti
on a
nd f
e
a
tu
r
e
r
e
s
a
m
pl
in
g i
n C
A
D
p
r
e
di
c
ti
ons
. F
e
a
tu
r
e
s
e
le
c
ti
on i
s
u
s
e
d t
o f
in
d t
he
m
os
t
r
e
le
va
nt
f
e
a
tu
r
e
s
to
C
A
D
pr
e
di
c
ti
on
s
.
W
hi
le
h
a
ndl
in
g
i
m
ba
la
nc
e
d
da
ta
,
w
e
r
e
vi
e
w
e
d
s
e
ve
r
a
l
f
e
a
tu
r
e
r
e
s
a
m
pl
in
g.
W
e
us
e
S
V
M
w
it
h
hype
r
pa
r
a
m
e
t
e
r
tu
ni
ng
to
f
in
d
th
e
c
o
m
bi
na
ti
on
of
pa
r
a
m
e
te
r
s
to
m
a
ke
a
n
e
f
f
e
c
ti
ve
C
A
D
pr
e
di
c
ti
on.
2.
R
E
S
E
A
R
C
H
M
E
T
H
O
D
T
he
r
e
a
r
e
f
our
m
a
in
s
te
ps
to
c
om
pl
e
te
th
is
r
e
s
e
a
r
c
h,
a
s
s
how
n
in
F
ig
u
r
e
1
.
T
he
f
ir
s
t
s
te
p
is
da
ta
e
xpl
or
a
ti
on,
f
ol
lo
w
e
d
by
da
ta
pr
e
pr
oc
e
s
s
in
g.
N
e
xt
,
w
e
u
s
e
f
e
a
t
ur
e
s
e
le
c
ti
on
to
de
te
r
m
in
e
w
hi
c
h
f
e
a
tu
r
e
s
ha
ve
th
e
m
os
t
im
por
ta
nc
e
on
th
e
t
a
r
ge
t
va
r
ia
bl
e
.
A
f
te
r
id
e
nt
if
yi
ng
th
e
r
e
le
va
nt
f
e
a
tu
r
e
s
,
th
e
da
ta
s
e
t
is
di
vi
de
d
in
to
tr
a
in
in
g
a
nd
te
s
ti
ng
s
e
ts
f
or
th
e
pur
pos
e
of
im
pl
e
m
e
nt
in
g
m
u
lt
i
pl
e
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
s
.
T
he
la
s
t
s
te
p
is
m
ode
l
e
va
lu
a
ti
on. T
hi
s
s
e
c
ti
on dis
c
us
s
e
s
t
h
e
pr
oc
e
s
s
e
s
a
nd pr
oc
e
dur
e
s
i
nvol
ve
d i
n doing t
hi
s
r
e
s
e
a
r
c
h.
2.1. Dat
as
e
t
d
e
s
c
r
ip
t
io
n
W
e
us
e
th
e
Z
-
A
li
z
a
de
h
s
a
ni
da
ta
s
e
t
dow
nl
oa
de
d
f
r
om
th
e
U
C
I
m
a
c
hi
ne
le
a
r
ni
ng
r
e
pos
it
or
y.
T
he
da
ta
s
e
t
c
ont
a
in
s
r
e
c
or
ds
of
303
pa
ti
e
nt
s
w
ho
vi
s
it
e
d
th
e
S
ha
he
e
d
R
a
ja
e
i
C
a
r
di
ova
s
c
ul
a
r
,
M
e
di
c
a
l,
a
nd
R
e
s
e
a
r
c
h
C
e
nt
e
r
in
I
r
a
n.
E
a
c
h
pa
ti
e
nt
ha
s
54
f
e
a
tu
r
e
s
to
di
a
gno
s
e
C
A
D
.
T
he
s
e
f
e
a
tu
r
e
s
a
r
e
gr
oupe
d
in
to
f
our
c
a
te
gor
ie
s
:
de
m
ogr
a
phi
c
,
s
ym
pt
om
a
nd
e
xa
m
in
a
ti
on,
e
le
c
tr
o
c
a
r
di
ogr
a
m
(
E
C
G
)
,
a
nd
la
bor
a
to
r
y
a
nd
e
c
ho
f
e
a
tu
r
e
s
.
P
a
ti
e
nt
s
a
r
e
c
a
te
gor
iz
e
d
a
s
ha
vi
ng
C
A
D
if
th
e
y
e
xp
e
r
ie
nc
e
s
te
nos
is
in
one
of
th
e
ir
c
or
ona
r
y
a
r
te
r
ie
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ff
e
c
ti
v
e
pr
e
di
c
ti
v
e
m
od
e
ll
in
g f
or
c
or
ona
r
y
ar
te
r
y
di
s
e
as
e
s
us
in
g s
uppor
t
…
(
K
unc
ah
y
o Se
ty
o N
ug
r
oho)
347
m
or
e
th
a
n
or
e
qua
l
to
50
%
.
A
to
ta
l
o
f
216
pa
ti
e
nt
s
in
th
e
da
ta
s
e
t
ha
d
th
is
di
s
e
a
s
e
,
w
hi
le
th
e
r
e
s
t
w
e
r
e
nor
m
a
l
pa
ti
e
nt
s
.
T
hi
s
s
how
s
th
a
t
th
e
da
ta
s
e
t
h
a
s
a
n
unb
a
la
nc
e
d
c
la
s
s
di
s
tr
ib
ut
io
n.
T
he
t
a
r
ge
t
f
e
a
tu
r
e
on
th
e
d
a
ta
s
e
t
is
c
a
th
w
it
h a
C
A
D
va
lu
e
f
or
pa
ti
e
nt
s
w
it
h
c
or
ona
r
y a
r
te
r
y di
s
e
a
s
e
a
nd
n
or
m
a
l
f
or
nor
m
a
l
pa
ti
e
nt
s
.
F
ig
ur
e
1
. R
e
s
e
a
r
c
h m
e
th
od de
s
ig
n f
or
C
A
D
pr
e
di
c
ti
on
2.2.
D
at
a
e
xp
lo
r
at
io
n
O
ur
da
ta
s
e
t
ha
s
m
a
ny
di
ve
r
s
e
f
e
a
tu
r
e
s
,
s
o
th
is
s
te
p
is
ta
ke
n
to
e
xpl
or
e
th
e
da
ta
s
e
t
to
ge
t
u
s
e
f
ul
in
s
ig
ht
s
th
r
ough
vi
s
ua
li
z
a
ti
on
a
nd
da
ta
a
n
a
ly
s
is
.
T
hi
s
s
te
p
a
l
s
o
he
lp
s
us
f
in
d
out
th
e
m
is
s
in
g
va
lu
e
s
a
nd
id
e
nt
if
y t
he
t
ype
s
of
nume
r
ic
f
e
a
tu
r
e
s
a
nd c
a
te
gor
ic
a
l
f
e
a
tu
r
e
s
i
n t
he
da
ta
s
e
t.
2.
3
. D
at
a p
r
e
p
r
oc
e
s
s
in
g
R
e
a
l
-
w
or
d
da
ta
s
e
ts
h
a
ve
in
c
om
pl
e
te
,
in
c
on
s
is
te
nt
,
a
nd
e
ve
n
h
a
ve
m
is
s
in
g
va
lu
e
on
s
pe
c
if
ic
f
e
a
tu
r
e
s
.
D
a
ta
pr
e
pr
oc
e
s
s
in
g
us
e
d
to
c
le
a
n
a
nd
f
or
m
a
t
th
e
r
a
w
da
ta
in
th
e
da
ta
s
e
t
s
o
th
a
t
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
s
c
a
n
e
a
s
il
y
r
e
pr
e
s
e
nt
th
e
f
e
a
tu
r
e
s
e
t.
I
n
th
is
s
tu
dy,
w
e
im
pl
e
m
e
n
te
d
s
e
ve
r
a
l
da
ta
pr
e
pr
oc
e
s
s
in
g
s
te
p
s
.
T
he
f
ir
s
t
s
te
p
is
to
c
onve
r
t
c
a
te
gor
ic
a
l
f
e
a
tu
r
e
s
to
num
e
r
ic
va
lu
e
s
be
c
a
u
s
e
m
a
c
hi
n
e
le
a
r
ni
ng
a
lg
or
it
hm
s
c
a
n
onl
y
r
e
a
d
a
nd
pr
oc
e
s
s
num
e
r
ic
va
lu
e
s
.
N
e
xt
,
w
e
c
r
e
a
te
a
f
e
a
tu
r
e
m
a
tr
ix
t
ha
t
is
us
e
d
a
s
th
e
in
put
v
a
r
ia
bl
e
a
nd
th
e
ta
r
ge
t
va
r
ia
bl
e
. T
he
i
nput
f
e
a
tu
r
e
i
s
s
to
r
e
d i
nt
o
va
r
ia
bl
e
w
hi
le
t
he
t
a
r
g
e
t
f
e
a
tu
r
e
i
s
s
to
r
e
d i
nt
o
va
r
ia
bl
e
. T
he
f
in
a
l
s
te
p
is
nor
m
a
li
z
in
g
th
e
da
ta
to
r
e
s
c
a
le
th
e
num
e
r
ic
f
e
a
tu
r
e
s
in
to
r
a
nge
s
0
a
nd
1
us
e
d
a
m
in
-
m
a
x
s
c
a
le
r
,
a
s
s
how
n i
n (
1)
.
′
=
−
min
(
)
ma
x
(
)
−
min
(
)
(
1)
2.
4
. F
e
at
u
r
e
s
e
le
c
t
io
n
T
he
f
e
a
tu
r
e
s
u
s
e
d
to
tr
a
in
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
s
ha
ve
a
s
ig
ni
f
ic
a
nt
im
pa
c
t
on
th
e
pe
r
f
or
m
a
nc
e
of
th
e
f
in
a
l
m
ode
l.
I
r
r
e
le
va
nt
f
e
a
tu
r
e
s
c
a
n
ha
ve
a
ne
ga
ti
ve
im
pa
c
t
on
th
e
r
e
s
ul
ti
ng
m
ode
l
[
26]
.
T
o
id
e
nt
if
y
f
e
a
tu
r
e
s
th
a
t
a
f
f
e
c
t
th
e
t
a
r
ge
t
va
r
ia
bl
e
,
f
e
a
tu
r
e
s
e
le
c
ti
on
c
a
n
be
us
e
d.
F
e
a
tu
r
e
s
e
le
c
ti
on
is
th
e
pr
oc
e
s
s
of
r
e
duc
in
g
th
e
num
be
r
of
f
e
a
tu
r
e
s
in
a
da
ta
s
e
t
in
or
de
r
to
im
p
r
ove
th
e
m
ode
l'
s
pe
r
f
or
m
a
nc
e
[
27]
.
W
e
us
e
d
f
e
a
tu
r
e
s
e
le
c
ti
on
to
pr
e
di
c
t
w
hi
c
h
f
e
a
tu
r
e
s
w
e
r
e
m
os
t
im
por
t
a
nt
in
in
f
lu
e
nc
in
g
pa
ti
e
nt
s
w
it
h
C
A
D
or
not
.
e
xt
r
e
m
e
ly
r
a
ndomi
z
e
d
tr
e
e
s
c
la
s
s
if
ie
r
is
a
n
e
ns
e
m
bl
e
le
a
r
ni
ng
ty
pe
us
e
d
f
or
f
e
a
tu
r
e
s
e
le
c
ti
o
n.
I
n
th
is
m
e
th
od,
e
a
c
h
de
c
is
io
n
tr
e
e
is
ge
n
e
r
a
te
d
f
r
om
th
e
tr
a
in
in
g
s
a
m
pl
e
.
T
he
n,
a
t
e
a
c
h
te
s
t
node
,
th
e
de
c
i
s
io
n
tr
e
e
is
gi
ve
n
a
r
a
ndom
s
a
m
pl
e
of
k
f
e
a
tu
r
e
s
of
a
ll
f
e
a
tu
r
e
s
,
w
he
r
e
e
a
c
h
de
c
is
io
n
tr
e
e
m
us
t
c
hoo
s
e
th
e
be
s
t
f
e
a
tu
r
e
to
s
e
pa
r
a
te
th
e
da
ta
ba
s
e
d
on
th
e
G
in
i
I
nde
x
va
lu
e
.
T
hi
s
r
a
ndom
f
e
a
tu
r
e
w
il
l
pr
ovi
de
s
e
ve
r
a
l
unc
or
r
e
la
te
d
de
c
is
io
n
tr
e
e
s
.
T
hi
s
va
lu
e
is
r
e
f
e
r
r
e
d
to
a
s
th
e
f
e
a
tu
r
e
'
s
G
in
i
I
m
por
ta
nc
e
.
T
o
m
a
ke
th
e
f
e
a
tu
r
e
s
e
le
c
ti
on
pr
oc
e
s
s
e
a
s
i
e
r
,
e
a
c
h
f
e
a
tu
r
e
i
s
gr
a
de
d a
c
c
or
di
ng t
o i
ts
G
in
i
I
m
por
ta
nc
e.
2.
5
.
D
at
a
s
e
p
ar
at
io
n
D
a
ta
s
e
pa
r
a
ti
on
is
us
e
d
to
e
va
lu
a
t
e
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
s
'
pe
r
f
or
m
a
nc
e
w
he
n
pr
e
di
c
ti
ng
da
ta
th
a
t
w
a
s
not
us
e
d
to
tr
a
in
th
e
m
ode
l.
D
iv
id
e
th
e
d
a
ta
s
e
t
in
to
t
w
o
s
ubs
e
ts
u
s
in
g
th
e
da
ta
s
e
pa
r
a
ti
on
pr
oc
e
s
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
11
, N
o.
1
,
M
a
r
c
h
20
22
:
345
-
355
348
T
he
f
ir
s
t
s
ubs
e
t
is
ut
i
li
z
e
d
to
tr
a
in
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
s
in
or
de
r
to
ge
ne
r
a
te
pr
e
di
c
ti
on
m
ode
ls
.
T
he
s
e
c
ond s
ub
s
e
t
is
t
he
t
e
s
t
s
e
t
on w
hi
c
h t
he
pr
e
di
c
ti
on mode
l
is
e
v
a
lu
a
te
d. W
e
t
r
a
in
e
d on 75%
of
t
he
da
ta
s
e
t
a
nd
te
s
te
d t
he
m
ode
l
on t
he
r
e
m
a
in
in
g 25%
.
2.
6
.
S
t
r
at
if
ie
d
k
-
f
ol
d
W
he
n
pe
r
f
or
m
in
g
th
e
da
ta
s
e
pa
r
a
ti
on
pr
oc
e
dur
e
,
th
e
m
a
in
pr
o
bl
e
m
m
us
t
be
e
nough
da
ta
to
di
vi
de
th
e
da
ta
s
e
t
in
to
tr
a
in
in
g
da
ta
a
nd
te
s
t
da
ta
a
s
da
ta
r
e
pr
e
s
e
nt
a
ti
ons
f
ol
lo
w
in
g
th
e
p
r
obl
e
m
dom
a
in
.
T
he
r
e
f
or
e
,
th
is
pr
oc
e
dur
e
is
not
s
ui
ta
bl
e
f
or
e
va
lu
a
ti
ng
m
ode
l
p
e
r
f
or
m
a
nc
e
if
th
e
r
e
a
r
e
f
e
w
da
ta
s
e
ts
a
va
il
a
bl
e
. T
he
r
e
w
il
l
not
be
e
nough da
ta
on t
he
t
r
a
in
in
g o
r
t
e
s
ti
ng s
ubs
e
t
f
or
t
he
m
od
e
l
to
l
e
a
r
n t
he
e
f
f
e
c
ti
ve
m
a
ppi
ng f
r
om
i
nput
t
o
out
put
. P
r
e
di
c
ti
on pe
r
f
or
m
a
nc
e
c
a
n be
t
oo opti
m
is
ti
c
(
good pr
e
di
c
t
io
n)
or
t
oo pe
s
s
im
is
ti
c
(
ba
d)
.
A
n
a
lt
e
r
na
ti
ve
m
e
th
od
th
a
t
c
a
n
be
u
s
e
d
if
do
not
ha
ve
e
nou
gh
da
ta
is
th
e
K
-
F
ol
d
pr
oc
e
dur
e
by
f
ol
di
ng
K
a
s
m
uc
h
da
ta
a
nd
r
e
pe
a
ti
ng
th
e
pr
oc
e
s
s
a
s
m
a
ny
a
s
K
a
s
w
e
ll
.
O
ne
ty
pe
of
th
is
pr
oc
e
dur
e
is
a
S
tr
a
ti
f
ie
d
K
-
f
ol
d,
a
s
s
how
n
in
F
ig
ur
e
2
.
S
tr
a
ti
f
ie
d
K
-
F
ol
d
is
he
lp
f
ul
if
th
e
a
va
il
a
bl
e
d
a
ta
s
e
t
is
f
e
w
a
nd
ha
s
a
n
unba
la
nc
e
d
c
la
s
s
di
s
tr
ib
ut
io
n.
W
e
w
a
nt
to
m
a
in
ta
in
th
e
c
la
s
s
i
m
ba
la
nc
e
to
r
e
pr
e
s
e
nt
s
om
e
in
f
or
m
a
ti
on
a
bout
w
ha
t
th
e
m
ode
l
is
tr
yi
ng
to
p
r
e
di
c
t.
I
n
th
is
s
tu
dy,
w
e
us
e
a
c
om
bi
na
ti
on
of
th
e
S
t
r
a
ti
f
ie
d
K
-
F
ol
d
pr
oc
e
dur
e
t
o
c
onduc
t
a
f
in
a
l
e
va
lu
a
ti
on
of
th
e
pe
r
f
or
m
a
nc
e
of
th
e
im
pl
e
m
e
nt
e
d
m
ode
l.
A
f
te
r
s
e
p
a
r
a
te
d
th
e
tr
a
in
in
g
a
nd
te
s
ti
ng
s
e
t
in
th
e
pr
e
vi
ous
s
te
ps
,
w
e
f
ur
th
e
r
di
vi
de
d
th
e
tr
a
in
in
g
s
e
t
in
to
va
li
da
ti
on
s
e
t
to
va
li
da
t
e
t
he
m
a
c
hi
ne
le
a
r
ni
ng a
lg
or
it
hm
s
pe
r
f
or
m
a
nc
e
dur
in
g
th
e
k
-
f
ol
d
i
te
r
a
ti
on pr
o
c
e
s
s
. W
e
a
ls
o pe
r
f
or
m
e
d f
e
a
tu
r
e
r
e
s
a
m
pl
in
g t
o
ba
la
nc
e
th
e
di
s
tr
ib
ut
io
n
of
c
la
s
s
e
s
in
th
e
tr
a
in
in
g
s
e
t
dur
in
g
th
is
pr
oc
e
s
s
.
T
he
ge
ne
r
a
te
d
pr
e
di
c
ti
on
m
ode
l
is
f
in
a
ll
y t
e
s
te
d us
in
g t
he
te
s
ti
ng
s
e
t
a
s
t
he
f
in
a
l
r
e
s
ul
t
of
t
he
pr
e
di
c
te
d pe
r
f
or
m
a
nc
e
.
F
ig
ur
e
2.
S
tr
a
ti
f
ie
d
k
-
f
ol
d s
c
he
m
a
2.
7
. F
e
at
u
r
e
r
e
s
am
p
li
n
g
I
m
ba
la
nc
e
da
ta
c
a
us
e
s
th
e
m
ode
l
to
be
bi
a
s
e
d
in
c
hoos
in
g
th
e
m
a
jo
r
it
y
c
la
s
s
.
T
he
r
e
a
r
e
m
a
ny
w
a
y
s
to
ha
ndl
e
a
d
a
ta
s
e
t
w
it
h
a
n
unba
l
a
nc
e
d
c
la
s
s
di
s
tr
ib
ut
io
n,
in
c
lu
di
ng
c
ol
le
c
ti
ng
m
or
e
da
ta
,
tr
yi
ng
va
r
ia
ti
ons
in
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
s
,
us
in
g
bot
h
ove
r
s
a
m
pl
in
g
a
nd
un
de
r
s
a
m
pl
in
g
te
c
hni
que
s
.
C
ol
le
c
ti
ng
m
or
e
da
ta
is
im
pos
s
ib
le
be
c
a
us
e
it
r
e
qui
r
e
s
m
or
e
ti
m
e
a
nd
c
o
s
ts
,
w
hi
le
u
nde
r
s
a
m
pl
in
g
te
c
hni
que
s
c
a
n
c
a
u
s
e
th
e
lo
s
s
of
im
por
ta
nt
in
f
or
m
a
ti
on
in
th
e
da
ta
s
e
t.
T
he
r
e
f
or
e
,
in
th
is
s
tu
dy,
w
e
f
oc
us
on
ove
r
s
a
m
pl
in
g
te
c
hni
que
s
S
M
O
T
E
a
nd
A
D
A
S
Y
N
s
o
th
a
t
w
e
hope
not
to
lo
s
e
a
ny
in
f
or
m
a
ti
on
t
ha
t
m
ig
ht
be
va
lu
a
bl
e
in
th
e
da
ta
s
e
t.
S
M
O
T
E
us
e
s
a
k
-
ne
a
r
e
s
t
ne
ig
hbor
s
(
k
-
NN
)
-
ba
s
e
d
di
s
ta
nc
e
a
ppr
oa
c
h
t
o
c
r
e
a
te
s
ynt
he
ti
c
da
ta
[
28]
.
F
ir
s
t,
th
e
da
ta
is
r
a
ndoml
y
s
e
le
c
te
d
f
r
om
th
e
m
in
or
it
y
c
la
s
s
,
th
e
n
K
is
th
e
c
lo
s
e
s
t
ne
ig
hbor
of
th
e
da
ta
.
S
ynt
he
ti
c
da
ta
is
ge
ne
r
a
te
d
be
twe
e
n
r
a
ndoml
y
s
e
le
c
t
e
d
a
nd
K
-
ne
a
r
e
s
t
da
ta
.
T
hi
s
s
te
p
i
s
r
e
pe
a
te
d
unt
il
th
e
m
in
or
it
y
c
la
s
s
ha
s
th
e
s
a
m
e
pr
opor
ti
on
a
s
th
e
m
a
jo
r
it
y
c
la
s
s
.
M
e
a
nw
hi
le
,
A
D
A
S
Y
N
is
a
va
r
ia
ti
on
of
S
M
O
T
E
by
c
r
e
a
ti
ng
s
ynt
he
ti
c
da
ta
ba
s
e
d
on
da
ta
de
ns
it
y
[
29]
.
T
he
s
ynt
h
e
ti
c
da
ta
g
e
ne
r
a
te
d
w
il
l
be
in
v
e
r
s
e
ly
pr
opor
ti
ona
l
to
th
e
de
ns
it
y
of
th
e
m
in
or
it
y
c
la
s
s
.
T
ha
t
i
s
,
m
or
e
s
ynt
he
ti
c
da
ta
is
g
e
ne
r
a
te
d
in
th
e
f
e
a
tu
r
e
s
pa
c
e
w
he
r
e
th
e
d
e
ns
it
y
of
t
he
m
in
or
it
y c
la
s
s
i
s
l
ow
, a
nd l
e
s
s
or
e
ve
n l
e
s
s
s
ynt
he
ti
c
da
ta
is
ge
ne
r
a
te
d i
n t
he
hi
gh de
ns
it
y m
in
or
it
y c
la
s
s
[
21]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ff
e
c
ti
v
e
pr
e
di
c
ti
v
e
m
od
e
ll
in
g f
or
c
or
ona
r
y
ar
te
r
y
di
s
e
as
e
s
us
in
g s
uppor
t
…
(
K
unc
ah
y
o Se
ty
o N
ug
r
oho)
349
2.
8
.
S
u
p
p
or
t
ve
c
t
or
m
ac
h
in
e
T
he
S
V
M
is
a
hi
ghl
y
e
f
f
e
c
ti
ve
c
la
s
s
if
ic
a
ti
on
a
lg
or
it
hm
by
f
i
ndi
ng
de
c
is
io
n
bounda
r
ie
s
known
a
s
hype
r
pl
a
ne
s
.
T
h
e
opt
im
a
l
hyp
e
r
pl
a
ne
s
e
pa
r
a
te
s
th
e
in
s
ta
nc
e
s
c
or
r
e
c
tl
y
in
to
e
a
c
h
c
la
s
s
.
T
he
m
a
r
gi
ns
on
th
e
opt
im
a
l
hype
r
pl
a
ne
a
nd
in
s
ta
nc
e
s
in
tr
a
in
in
g
a
r
e
m
a
xi
m
iz
e
d
to
f
it
th
e
da
ta
.
T
he
S
V
M
m
ode
l
i
s
not
de
li
c
a
te
to
ot
he
r
in
f
or
m
a
ti
on
f
oc
us
e
s
.
I
ts
poi
nt
i
s
to
tr
a
c
k
dow
n
th
e
be
s
t
di
vi
s
io
n
li
ne
,
f
or
e
xa
m
pl
e
,
th
e
id
e
a
l
hype
r
pl
a
n
e
be
twe
e
n
th
e
two
c
la
s
s
e
s
of
te
s
ts
,
to
ha
ve
th
e
m
os
t
s
ig
ni
f
ic
a
nt
di
s
ta
nc
e
c
onc
e
iv
a
bl
e
to
e
ve
r
y
on
e
of
th
e
two
c
la
s
s
e
s
of
he
lp
ve
c
to
r
s
.
T
he
s
e
pa
r
a
to
r
li
ne
di
c
ta
te
s
th
e
in
di
c
a
to
r
in
c
lu
de
f
or
e
a
c
h
pr
e
s
c
ie
n
t
c
la
s
s
.
F
ig
ur
e
3
s
how
s
t
he
ve
c
to
r
m
a
c
hi
ne
i
n 2
-
di
m
e
ns
io
na
l
s
pa
c
e
[
16]
.
F
ig
ur
e
3
.
S
V
M
i
n 2
-
di
m
e
ns
io
na
l
s
pa
c
e
[
16]
A
hype
r
pl
a
ne
w
it
h
a
w
id
e
r
m
a
r
gi
n
is
pr
oj
e
c
te
d
to
be
m
o
r
e
a
c
c
ur
a
te
th
a
n
one
w
it
h
a
s
m
a
ll
e
r
m
a
r
gi
n
w
he
n
c
la
s
s
if
yi
ng
f
ut
ur
e
da
ta
.
T
h
e
r
e
f
or
e
,
th
e
hyp
e
r
pl
a
ne
w
it
h
th
e
la
r
ge
s
t
m
a
r
gi
n
w
il
l
be
s
e
a
r
c
he
d
f
or
.
T
he
f
unc
ti
on ha
s
t
he
f
ol
lo
w
in
g i
n (
2)
[
30]
.
(
)
=
[
∑
(
,
)
+
=
1
]
(
2)
H
ow
e
ve
r
,
in
(
2)
c
a
n
be
a
ppl
ie
d
if
th
e
s
a
m
pl
e
da
t
a
us
e
d
c
a
n
be
s
e
pa
r
a
te
d
li
ne
a
r
ly
.
K
e
r
ne
l
m
e
th
ods
e
na
bl
e
th
e
tr
a
ns
f
or
m
a
t
io
n
of
da
ta
in
to
huge
di
m
e
ns
io
ns
f
or
c
la
s
s
if
ic
a
ti
on
c
ha
ll
e
nge
s
.
A
s
is
th
e
c
a
s
e
w
it
h
da
t
a
s
a
m
pl
e
s
th
a
t
c
a
nnot
b
e
s
pl
it
li
ne
a
r
ly
,
th
e
k
e
r
ne
l
f
unc
ti
on
c
onv
e
r
ts
th
e
da
ta
to
a
hi
ghe
r
-
di
m
e
ns
io
na
l
s
pa
c
e
w
it
hout
a
c
tu
a
ll
y
c
ha
ngi
ng
it
to
th
a
t
s
p
a
c
e
.
In
(
3)
c
a
n
be
a
p
pl
ie
d
w
he
n
th
e
da
ta
s
a
m
pl
e
s
it
ua
ti
on
c
a
nnot
be
s
e
pa
r
a
te
d
li
ne
a
r
ly
.
(
)
=
[
∑
(
,
)
+
=
1
]
(
3)
T
he
ke
r
ne
l
f
unc
ti
on
(
,
)
is
e
qua
ls
to
(
,
)
a
nd
is
th
e
non
-
li
ne
a
r
s
pa
c
e
f
r
om
th
e
or
ig
in
a
l
s
pa
c
e
to
hi
gh
di
m
e
ns
io
na
l
s
pa
c
e
.
W
he
r
e
,
an
d
a
r
e
ke
r
ne
l
pa
r
a
m
e
te
r
s
,
a
nd
th
e
f
our
ba
s
ic
ke
r
ne
ls
a
r
e
gi
ve
n
a
s
f
ol
lo
w
s
in
(
4)
-
(
7
)
. I
n t
hi
s
s
tu
dy, we
us
e
a
ll
ke
r
ne
ls
t
o f
in
d t
he
be
s
t
S
V
M
pe
r
f
or
m
a
nc
e
.
:
(
,
)
=
(
4)
:
(
,
)
=
(
+
)
,
>
0
(
5)
(
)
:
(
,
)
=
(
−
‖
−
‖
2
)
,
>
0
(
6)
:
(
,
)
=
ℎ
(
+
)
(
7)
2.9.
E
val
u
at
io
n
m
e
t
r
ic
s
T
he
m
ode
l
ge
ne
r
a
te
d
dur
in
g
th
e
tr
a
in
in
g
pha
s
e
is
u
s
e
d
to
obt
a
in
pr
e
di
c
ti
ve
r
e
s
ul
ts
f
r
om
popula
ti
on
da
ta
.
I
n
th
e
c
la
s
s
if
ic
a
ti
on,
th
e
c
onf
us
io
n
m
a
tr
ix
de
s
c
r
ib
e
s
th
e
m
ode
l'
s
pe
r
f
or
m
a
nc
e
by
c
a
lc
ul
a
ti
ng
w
hi
c
h
c
la
s
s
e
s
a
r
e
pr
e
di
c
te
d
c
or
r
e
c
tl
y
a
nd
in
c
or
r
e
c
tl
y
a
nd
w
h
a
t
ty
p
e
s
of
e
r
r
or
s
a
r
e
m
a
de
.
T
r
ue
pos
it
iv
e
(
T
P
)
is
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
11
, N
o.
1
,
M
a
r
c
h
20
22
:
345
-
355
350
de
f
in
e
d
a
s
pos
it
iv
e
in
s
ta
nc
e
s
th
a
t
a
r
e
pr
e
di
c
te
d
to
be
t
r
ue
.
F
o
r
e
xa
m
pl
e
,
a
pa
ti
e
nt
w
it
h
C
A
D
is
pr
e
di
c
te
d
to
ha
ve
tr
ue
C
A
D
.
T
r
ue
n
e
ga
ti
ve
(
T
N
)
is
de
f
in
e
d
a
s
n
e
ga
ti
ve
in
s
ta
nc
e
s
th
a
t
a
r
e
pr
e
di
c
te
d
to
b
e
tr
ue
.
F
or
e
xa
m
pl
e
,
a
pa
ti
e
nt
w
ho
doe
s
not
ha
ve
C
A
D
is
pr
e
di
c
te
d
no
t
t
o
ha
ve
C
A
D
.
F
a
ls
e
pos
it
iv
e
(
F
P
)
is
ne
ga
ti
ve
in
s
ta
nc
e
s
th
a
t
a
r
e
pr
e
di
c
te
d a
s
po
s
it
iv
e
in
s
t
a
nc
e
s
.
F
or
e
xa
m
pl
e
,
a
pa
ti
e
nt
w
ho
doe
s
not
ha
v
e
C
A
D
is
pr
e
di
c
te
d
to
ha
ve
C
A
D
.
F
a
ls
e
ne
ga
ti
ve
(
F
N
)
is
po
s
it
iv
e
in
s
ta
nc
e
s
th
a
t
a
r
e
pr
e
di
c
te
d
a
s
ne
g
a
ti
ve
in
s
ta
nc
e
s
.
F
or
e
xa
m
pl
e
,
a
pa
ti
e
nt
w
ho ha
s
C
A
D
i
s
pr
e
di
c
te
d not t
o ha
v
e
C
A
D
.
T
he
m
os
t
f
r
e
que
nt
ly
u
s
e
d
pe
r
f
or
m
a
nc
e
m
e
tr
ic
b
a
s
e
d
on
th
e
c
onf
us
io
n
m
a
tr
ix
f
or
c
la
s
s
if
ic
a
ti
on
is
a
c
c
ur
a
c
y
.
A
c
c
ur
a
c
y
is
th
e
r
a
ti
o
of
tr
ue
pr
e
di
c
ti
ons
(
T
P
a
nd
T
N
)
w
it
h
th
e
ove
r
a
ll
da
ta
th
a
t
de
s
c
r
ib
e
s
th
e
le
ve
l
of
c
lo
s
e
ne
s
s
of
th
e
pr
e
di
c
te
d
va
lu
e
to
th
e
a
c
tu
a
l
va
lu
e
,
a
s
s
how
n
in
(
8)
.
I
n
th
e
tr
a
in
in
g
pha
s
e
,
th
e
m
ode
l’
s
a
c
c
ur
a
c
y
is
o
bt
a
in
e
d
f
r
om
th
e
a
ve
r
a
ge
of
e
a
c
h
f
ol
d
in
th
e
c
r
os
s
-
va
li
da
ti
on.
T
he
s
ta
nda
r
d
de
vi
a
ti
on
w
a
s
a
ls
o
c
a
lc
ul
a
te
d t
o s
e
e
t
he
va
r
ia
nc
e
. T
h
e
pr
obl
e
m
w
it
h unba
la
nc
e
d da
ta
i
s
ne
ga
ti
ve
i
ns
ta
nc
e
s
w
it
h t
he
m
a
jo
r
it
y c
la
s
s
a
nd
pos
it
iv
e
in
s
ta
n
c
e
s
w
it
h
f
e
w
e
r
c
la
s
s
e
s
.
T
o
in
te
r
pr
e
ti
ng
th
e
m
ode
l
pe
r
f
or
m
a
nc
e
w
it
h
unba
la
nc
e
d
da
ta
,
r
e
c
e
iv
e
r
ope
r
a
ti
ng c
ha
r
a
c
te
r
is
ti
c
(
R
O
C
)
c
ur
ve
a
r
e
us
e
d. T
he
R
O
C
c
ur
ve
i
s
obt
a
in
e
d f
r
om
t
he
t
r
ue
pos
it
iv
e
r
a
te
(
T
P
R
)
a
s
i
n (
9)
a
nd t
he
f
a
ls
e
pos
it
iv
e
r
a
te
(
F
P
R
)
a
s
i
n (
10)
.
=
+
+
+
+
(
8)
=
+
(
9)
=
+
(
10)
3.
R
E
S
U
L
T
S
A
N
D
D
I
S
C
U
S
S
I
O
N
B
a
s
e
d
on
th
e
r
e
s
e
a
r
c
h
f
r
a
m
e
w
or
k
in
F
ig
ur
e
1
,
th
e
f
ir
s
t
s
te
p
w
e
t
a
ke
is
pr
e
pr
oc
e
s
s
in
g
by
c
ha
ngi
ng
a
ll
c
a
te
gor
ic
a
l
f
e
a
tu
r
e
s
i
n t
he
da
ta
s
e
t
to
nume
r
ic
va
lu
e
s
a
nd
nor
m
a
li
z
in
g t
he
m
us
in
g t
he
m
in
m
a
x s
c
a
le
r
. N
e
xt
,
w
e
us
e
f
e
a
tu
r
e
s
e
le
c
ti
on
to
de
te
r
m
in
e
w
hi
c
h
f
e
a
tu
r
e
s
a
r
e
s
ig
ni
f
ic
a
n
t
a
nd
ha
ve
a
n
e
f
f
e
c
t
on
th
e
ta
r
ge
t
v
a
r
ia
bl
e
.
W
e
c
hoos
e
f
e
a
tu
r
e
s
u
s
in
g
th
e
a
ddi
ti
ona
l
tr
e
e
s
c
la
s
s
if
ie
r
.
F
r
om
th
e
r
e
s
ul
ts
of
f
e
a
tu
r
e
s
e
le
c
ti
on
,
w
e
f
ound
th
a
t
th
e
r
e
a
r
e
16
f
e
a
tu
r
e
s
th
a
t
a
r
e
c
or
r
e
la
te
d
w
it
h
th
e
ta
r
ge
t
v
a
r
ia
bl
e
s
s
h
ow
n
in
T
a
bl
e
1
.
I
n
e
a
c
h
s
ub
s
e
que
nt
te
s
t
,
w
e
c
om
pa
r
e
t
he
m
ode
l'
s
pe
r
f
or
m
a
nc
e
us
in
g a
ll
th
e
f
e
a
tu
r
e
s
a
nd f
e
a
tu
r
e
s
e
le
c
ti
on r
e
s
ul
ts
. W
e
w
a
nt
e
d t
o f
in
d out
i
f
f
e
a
tu
r
e
s
e
le
c
ti
on c
oul
d i
m
pr
ove
t
he
pe
r
f
or
m
a
nc
e
of
a
gi
ve
n m
o
de
l.
T
a
bl
e
1
.
T
he
f
e
a
tu
r
e
s
u
s
e
d a
r
e
ba
s
e
d on the
s
e
le
c
ti
on r
e
s
ul
ts
F
e
a
ut
ur
e
na
m
e
F
e
a
t
ur
e
c
a
t
e
gor
y
A
ge
D
e
m
ogr
a
phi
c
W
e
i
ght
D
e
m
ogr
a
phi
c
B
M
I
D
e
m
ogr
a
phi
c
H
T
N
D
e
m
ogr
a
phi
c
BP
S
ym
pt
om
a
nd e
xa
m
i
na
t
i
on
T
ypi
c
a
l
c
he
s
t
pa
i
n
S
ym
pt
om
a
nd e
xa
m
i
na
t
i
on
A
t
ypi
c
a
l
S
ym
pt
om
a
nd e
xa
m
i
na
t
i
on
N
ona
ngi
na
l
S
ym
pt
om
a
nd e
xa
m
i
na
t
i
on
T
i
nve
r
s
i
on
E
C
G
F
B
S
L
a
bor
a
t
or
y a
nd e
c
ho
TG
L
a
bor
a
t
or
y a
nd e
c
ho
E
S
R
L
a
bor
a
t
or
y a
nd e
c
ho
N
e
ut
L
a
bor
a
t
or
y a
nd
e
c
ho
EF
-
T
T
E
L
a
bor
a
t
or
y a
nd e
c
ho
R
e
gi
on R
M
W
A
L
a
bor
a
t
or
y a
nd e
c
ho
T
o
e
va
lu
a
t
e
th
e
pr
e
di
c
ti
ve
m
ode
l,
w
e
di
vi
de
d
th
e
da
ta
s
e
t
in
to
75%
f
or
th
e
tr
a
in
in
g
s
e
t
a
nd
th
e
r
e
m
a
in
de
r
f
or
th
e
te
s
t
s
e
t.
N
e
xt
,
th
e
10
-
la
ye
r
c
r
os
s
-
va
li
da
ti
on
p
r
oc
e
dur
e
w
e
a
ppl
ie
d
to
th
e
tr
a
in
in
g
s
e
t.
I
n
th
is
pr
oc
e
dur
e
,
th
e
tr
a
in
in
g
s
e
t
i
s
r
a
ndoml
y
di
vi
de
d
in
to
te
n
s
e
c
ti
ons
,
9
f
ol
ds
a
r
e
us
e
d
to
tr
a
in
th
e
c
la
s
s
if
ic
a
ti
on
m
ode
l
a
nd
th
e
r
e
m
a
in
in
g
1
f
ol
d
is
di
v
id
e
d
to
va
li
da
te
th
e
m
od
e
l.
T
hi
s
pr
oc
e
dur
e
is
pe
r
f
or
m
e
d
10
f
ol
ds
.
T
he
da
ta
s
e
t
c
ha
r
a
c
te
r
ha
s
c
la
s
s
i
m
ba
la
nc
e
,
th
e
va
lu
e
of
c
r
os
s
va
li
da
t
io
n i
s
a
m
e
tr
ic
t
ha
t
c
a
n be
us
e
d t
o e
va
lu
a
te
t
he
m
ode
l
be
c
a
us
e
th
e
f
ol
ds
a
r
e
m
a
de
w
it
h
th
e
s
a
m
e
num
be
r
of
s
a
m
pl
e
s
f
or
e
a
c
h
c
la
s
s
s
o
th
a
t
th
e
c
la
s
s
di
s
tr
ib
ut
io
n
c
a
n
be
opt
im
a
ll
y
ba
la
nc
e
d.
T
hi
s
pr
oc
e
dur
e
c
a
n
ul
ti
m
a
te
ly
pr
ovi
de
s
uf
f
ic
ie
nt
r
e
pr
e
s
e
nt
a
ti
on
of
th
e
m
in
or
it
y a
nd ma
jo
r
it
y c
la
s
s
e
s
i
n e
a
c
h gr
oup.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ff
e
c
ti
v
e
pr
e
di
c
ti
v
e
m
od
e
ll
in
g f
or
c
or
ona
r
y
ar
te
r
y
di
s
e
as
e
s
us
in
g s
uppor
t
…
(
K
unc
ah
y
o Se
ty
o N
ug
r
oho)
351
W
e
c
om
pa
r
e
d
th
e
pe
r
f
or
m
a
nc
e
of
th
e
va
r
io
us
num
be
r
of
c
la
s
s
if
ic
a
ti
ons
in
C
A
D
pr
e
di
c
ti
ons
on
di
f
f
e
r
e
nt
unba
la
nc
e
d
a
nd
ba
la
nc
e
d
da
ta
s
e
ts
.
W
e
c
hoos
e
S
V
M
a
s
ou
r
m
a
in
m
ode
l.
W
e
c
om
pa
r
in
g
th
e
pe
r
f
or
m
a
nc
e
of
th
e
S
V
M
m
ode
l
w
it
h
s
e
v
e
r
a
l
ot
he
r
m
e
th
ods
s
uc
h
a
s
k
-
N
N
,
na
ïv
e
ba
ye
s
,
a
nd
de
c
is
io
n
tr
e
e
w
it
hout
a
ny
c
r
os
s
-
va
li
da
ti
on
pr
oc
e
dur
e
s
.
T
h
e
ne
xt
e
xp
e
r
im
e
nt
is
to
im
pl
e
m
e
nt
th
e
r
e
s
a
m
pl
in
g
f
e
a
tu
r
e
in
th
e
c
r
os
s
-
va
li
da
ti
on
pr
oc
e
dur
e
by
c
r
e
a
ti
ng
s
ynt
he
ti
c
da
ta
f
or
t
he
m
in
or
it
y
c
la
s
s
.
W
e
us
e
d
two
di
f
f
e
r
e
nt
ove
r
s
a
m
pl
in
g
te
c
hni
que
s
,
S
M
O
T
E
a
nd
A
D
A
S
Y
N
.
U
nba
la
nc
e
d
tr
a
in
in
g
da
ta
s
how
n
in
F
ig
ur
e
4
,
w
hi
le
th
e
ba
la
nc
e
d t
r
a
in
in
g da
ta
r
e
s
ul
ts
a
r
e
s
how
n i
n
F
ig
ur
e
s
5(
a
)
a
nd (
b)
.
F
ig
ur
e
1
. U
nba
la
nc
e
d on tr
a
in
in
g da
ta
(
a
)
(
b)
F
ig
ur
e
2
. T
he
r
e
s
ul
t
of
f
e
a
tu
r
e
r
e
s
a
m
pl
in
g
(
a
)
us
in
g S
M
O
T
E
a
n
d (
b)
us
in
g A
D
A
S
Y
N
T
he
m
a
jo
r
it
y
of
m
a
c
hi
ne
l
e
a
r
ni
ng
a
lg
or
it
hm
s
w
il
l
not
pr
oduc
e
opt
im
a
l
r
e
s
ul
ts
if
th
e
p
a
r
a
m
e
te
r
s
a
r
e
not
pr
ope
r
ly
s
pe
c
if
ie
d.
I
n
or
de
r
to
bui
ld
a
good
c
la
s
s
if
ic
a
ti
on
m
ode
l,
it
is
ve
r
y
im
por
ta
nt
to
s
e
le
c
t
th
e
pa
r
a
m
e
te
r
s
in
a
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
.
E
f
f
e
c
ti
ve
pa
r
a
m
e
te
r
in
it
ia
li
z
a
ti
on
is
s
it
ua
ti
on
-
de
pe
nde
nt
,
a
nd
e
a
c
h
s
it
ua
ti
on
m
a
y
r
e
qui
r
e
uni
que
pa
r
a
m
e
te
r
s
.
B
y
s
pe
c
if
yi
ng
th
e
a
ppr
opr
ia
te
pa
r
a
m
e
te
r
s
,
th
e
m
ode
l
c
a
n
be
gui
de
d
to
it
s
opt
im
a
l
s
ol
ut
io
n
[
31]
.
P
a
r
a
m
e
te
r
opt
im
iz
a
ti
on
is
ti
m
e
c
ons
um
in
g
if
done
m
a
nua
ll
y,
e
s
pe
c
i
a
ll
y
s
in
c
e
it
ha
s
m
a
ny
pa
r
a
m
e
te
r
s
.
T
he
bi
gge
s
t
pr
obl
e
m
in
s
e
tt
in
g
up
a
n
S
V
M
m
ode
l
is
c
hoos
in
g
ke
r
ne
l
f
unc
ti
ons
a
nd
th
e
ir
pa
r
a
m
e
te
r
va
lu
e
s
[
32]
,
[
33]
.
I
n
c
or
r
e
c
t
pa
r
a
m
e
te
r
s
e
tt
in
gs
le
a
d
to
poor
c
la
s
s
if
ic
a
ti
on
r
e
s
ul
ts
.
T
he
r
e
f
or
e
,
th
e
la
s
t
s
te
p
w
e
to
ok
w
a
s
a
gr
id
s
e
a
r
c
h
f
or
hype
r
-
pa
r
a
m
e
te
r
s
tu
ni
ng
to
f
in
d
th
e
opt
i
m
a
l
S
V
M
pa
r
a
m
e
te
r
.
H
ype
r
-
pa
r
a
m
e
te
r
s
a
r
e
de
f
in
e
d
us
in
g
th
e
m
in
im
um
va
lu
e
,
m
a
xi
m
um
va
lu
e
,
a
nd
th
e
n
um
be
r
of
s
te
ps
.
T
he
pa
r
a
m
e
t
e
r
s
w
e
a
r
e
lo
oki
ng
f
or
a
r
e
s
how
n
in
T
a
bl
e
2.
M
a
c
hi
ne
l
e
a
r
ni
ng
a
lg
or
it
hm
s
a
r
e
tr
a
in
e
d
on
im
ba
la
nc
e
d da
ta
us
in
g
a
ll
f
e
a
tu
r
e
s
a
nd f
r
om
t
he
f
e
a
tu
r
e
s
e
le
c
ti
o
n r
e
s
ul
ts
, a
s
s
how
n i
n T
a
bl
e
3.
B
a
s
e
d
on
T
a
bl
e
3,
S
V
M
s
how
s
th
a
t
it
ha
s
th
e
h
ig
he
s
t
a
c
c
ur
a
c
y
in
th
e
t
r
a
in
in
g
a
nd
te
s
ti
ng
pha
s
e
s
,
s
o
it
is
s
upe
r
io
r
to
a
ll
th
e
m
ode
ls
w
e
us
e
.
T
he
S
V
M
a
c
c
ur
a
c
y
obt
a
in
e
d
w
he
n
us
in
g
th
e
f
ul
l
f
e
a
tu
r
e
is
be
tt
e
r
th
a
n
us
in
g
f
e
a
tu
r
e
s
e
le
c
ti
on.
T
hi
s
is
c
on
s
is
te
nt
w
it
h
r
e
s
e
a
r
c
h
[
10]
th
a
t
S
V
M
pr
oduc
e
s
be
tt
e
r
a
c
c
ur
a
c
y
w
he
n
us
in
g
a
ll
f
e
a
tu
r
e
s
(
hi
gh
di
m
e
ns
io
na
l
d
a
ta
)
.
W
e
s
e
e
th
a
t
our
ba
s
e
m
ode
l
doe
s
a
r
e
a
s
ona
bl
e
jo
b
of
m
ode
li
ng
th
e
da
ta
.
B
ut
w
he
n
vi
e
w
e
d,
th
e
a
c
c
ur
a
c
y
va
lu
e
of
tr
a
in
in
g
a
nd
te
s
ti
ng
ha
s
qui
te
a
di
f
f
e
r
e
nc
e
.
W
e
w
a
nt
th
e
tr
a
in
in
g
va
lu
e
to
be
th
e
s
a
m
e
or
c
lo
s
e
to
th
e
te
s
t
va
lu
e
.
T
he
r
e
f
or
e
,
our
ne
xt
e
xpe
r
im
e
nt
us
e
s
s
tr
a
ti
f
ie
d
k
-
f
ol
d
to
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
11
, N
o.
1
,
M
a
r
c
h
20
22
:
345
-
355
352
va
li
da
te
th
e
m
ode
l.
B
a
s
e
d
on
T
a
bl
e
4
,
th
e
a
ve
r
a
g
e
a
c
c
ur
a
c
y
dur
in
g
th
e
tr
a
in
in
g
pha
s
e
de
c
r
e
a
s
e
s
f
or
th
e
w
hol
e
m
ode
l.
T
hi
s
s
how
s
th
a
t
our
m
ode
l
c
a
n
pr
e
di
c
t
th
e
te
s
t
da
ta
m
o
r
e
a
c
c
ur
a
te
ly
.
N
ow
w
e
ju
s
t
f
oc
us
onl
y
on
our
m
a
in
m
ode
l,
th
e
S
V
M
.
N
e
xt
,
w
e
w
il
l
pe
r
f
or
m
f
e
a
tu
r
e
r
e
s
a
m
pl
in
g
w
it
h
S
M
O
T
E
a
nd
A
D
A
S
Y
N
a
s
a
n e
f
f
or
t
to
ba
la
nc
e
th
e
tr
a
in
in
g
da
ta
dur
in
g
th
e
c
r
os
s
-
va
li
da
ti
on
pr
oc
e
dur
e
.
W
e
a
l
s
o
pe
r
f
or
m
m
ode
l
-
ba
s
e
d
pr
e
di
c
ti
ons
on
th
e
te
s
t
s
e
t
.
B
a
s
e
d
on
T
a
bl
e
5
,
th
e
S
V
M
m
ode
l
pr
ovi
de
s
good
p
e
r
f
or
m
a
nc
e
w
he
n
us
in
g
S
M
O
T
E
.
W
he
n
us
in
g
f
e
a
tu
r
e
s
e
le
c
ti
on,
tr
a
in
in
g
a
c
c
ur
a
c
y
dr
ops
w
hi
le
te
s
ti
ng
a
c
c
ur
a
c
y
goe
s
up.
T
hi
s
m
a
ke
s
it
pos
s
ib
le
f
or
our
m
ode
l
to
m
o
r
e
a
c
c
ur
a
te
ly
pr
e
di
c
t
t
e
s
t
d
a
ta
.
T
o
im
pr
ove
th
e
pe
r
f
or
m
a
nc
e
of
th
e
f
in
a
l
m
ode
l,
w
e
ta
ke
hype
r
pa
r
a
m
e
te
r
-
tu
ni
ng
to
f
in
d
th
e
m
os
t
opt
im
a
l
pa
r
a
m
e
te
r
s
in
th
e
S
V
M
us
in
g
f
e
a
tu
r
e
s
f
r
om
th
e
f
e
a
tu
r
e
s
e
le
c
ti
on
r
e
s
ul
ts
a
nd
S
M
O
T
E
in
s
tr
a
ti
f
ie
d
k
-
f
ol
d
.
B
a
s
e
d
on
gr
id
-
s
e
a
r
c
h,
t
he
be
s
t
S
V
M
m
ode
l
is
obt
a
in
e
d
us
in
g
th
e
pa
r
a
m
e
te
r
s
C
=
1000,
de
gr
e
e
=
1,
ga
m
m
a
=
0.001,
ke
r
ne
l:
R
B
F
.
B
y
us
in
g
c
r
os
s
-
va
li
da
ti
on,
th
e
hi
ghe
s
t
m
ode
l
a
c
c
ur
a
c
y
obt
a
in
e
d
in
th
e
te
s
t
s
e
t
r
e
a
c
h
e
d
8
8
%
,
a
s
s
how
n
in
T
a
bl
e
6
.
B
a
s
e
d
on
th
e
c
onf
us
io
n
m
a
tr
ix
in
F
ig
ur
e
3
,
th
e
m
ode
l
s
how
s
th
e
num
be
r
of
T
P
=
48
a
nd
T
N
=
19.
W
hi
le
th
e
R
O
C
c
ur
ve
in
F
ig
ur
e
4
s
how
s
th
a
t
th
e
m
ode
l
ha
s
good
pe
r
f
or
m
a
nc
e
be
c
a
us
e
th
e
c
ur
ve
is
a
w
a
y
f
r
om
th
e
ba
s
e
li
ne
to
th
e
T
P
R
a
xi
s
.
T
hi
s
m
e
a
n
s
th
a
t
th
e
m
ode
l
c
la
s
s
if
ie
s
m
or
e
da
ta
i
ns
ta
nc
e
s
c
or
r
e
c
tl
y.
T
a
bl
e
2
. G
r
id
s
e
a
r
c
h f
or
hype
r
-
pa
r
a
m
e
te
r
s
t
uni
ng t
o f
in
d
opt
im
a
l
S
V
M
pa
r
a
m
e
te
r
s
S
V
M
p
a
r
a
m
e
t
e
r
V
a
l
ue
r
a
nge
C
0.1, 1, 10, 100, 1000
G
a
m
m
a
1, 0.1, 0.001, 0.001, 0.0001
D
e
gr
e
e
1, 2, 3, 4, 5, 6
K
e
r
ne
l
L
i
ne
a
r
, P
ol
ynom
i
a
l
, R
B
F
, S
i
gm
oi
d
T
a
bl
e
3
.
T
r
a
in
in
g a
nd t
e
s
ti
ng s
c
or
e
s
e
ve
r
a
l
a
lg
or
it
hm
on t
he
i
m
ba
la
nc
e
d da
ta
s
e
t
M
ode
l
F
ul
l
F
e
a
t
ur
e
F
e
a
t
ur
e
F
e
a
t
ur
e
T
r
a
i
ni
ng
T
e
s
t
i
ng
T
r
a
i
ni
ng
T
e
s
t
i
ng
S
V
M
0.942
0.855
0.925
0.815
k
-
NN
0.885
0.776
0.881
0.868
N
a
ï
ve
B
a
ye
s
0.894
0.828
0.867
0.815
D
e
c
i
s
i
on T
r
e
e
1.000
0.723
1.000
0.776
T
a
bl
e
4
. A
ve
r
a
ge
c
r
os
s
-
va
li
da
ti
on s
c
or
e
t
r
a
in
e
d on the
i
m
ba
la
nc
e
d da
ta
s
e
t
M
ode
l
F
ul
l
f
e
a
t
ur
e
F
e
a
t
ur
e
f
e
a
t
ur
e
C
r
os
s
va
l
i
da
t
i
on s
c
or
e
S
V
M
0.872 ±
0.055
0.855 ±
0.072
k
-
NN
0.828 ±
0.037
0.842 ±
0.074
N
a
ï
ve
B
a
ye
s
0.854 ±
0.042
0.863 ±
0.059
D
e
c
i
s
i
on T
r
e
e
0.833 ±
0.041
0.797 ±
0.048
T
a
bl
e
5
. S
V
M
pe
r
f
or
m
a
nc
e
w
it
h f
e
a
tu
r
e
r
e
s
a
m
pl
in
g t
r
a
in
e
d on the
ba
la
nc
e
d da
ta
s
e
t
F
ul
l
F
e
a
t
ur
e
F
e
a
t
ur
e
S
e
l
e
c
t
i
on
T
r
a
i
ni
ng
C
r
os
s
V
a
l
i
da
t
i
on
T
e
s
t
i
ng
T
r
a
i
ni
ng
C
r
os
s
V
a
l
i
da
t
i
on
T
e
s
t
i
ng
S
M
O
T
E
0.942
0.859 ±
0.038
0.855
0.920
0.868 ±
0.091
0.881
A
D
A
S
Y
N
0.942
0.849 ±
0.091
0.881
0.876
0.828 ±
0.105
0.815
F
ig
ur
e
3
. C
onf
us
io
n m
a
tr
ix
of
S
V
M
ba
la
nc
e
d da
ta
s
e
t
w
it
h f
e
a
tu
r
e
s
e
le
c
ti
on
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ff
e
c
ti
v
e
pr
e
di
c
ti
v
e
m
od
e
ll
in
g f
or
c
or
ona
r
y
ar
te
r
y
di
s
e
as
e
s
us
in
g s
uppor
t
…
(
K
unc
ah
y
o Se
ty
o N
ug
r
oho)
353
T
a
bl
e
6
. H
ype
r
-
pa
r
a
m
e
te
r
t
uni
ng f
or
S
V
M
pe
r
f
or
m
a
nc
e
i
n ba
la
nc
e
d da
ta
s
e
t
w
it
h f
e
a
tu
r
e
s
e
le
c
ti
on
T
r
a
i
ni
ng
C
r
os
s
V
a
l
i
da
t
e
T
e
s
t
i
ng
0.925
0.877 ±
0.05
0.881
F
ig
ur
e
4
. R
O
C
c
ur
ve
of
S
V
M
ba
la
nc
e
d d
a
ta
s
e
t
w
it
h
f
e
a
tu
r
e
s
e
le
c
ti
on
4.
C
O
N
C
L
U
S
I
O
N
T
hi
s
s
tu
dy
ha
s
s
uc
c
e
e
de
d
in
m
a
ki
ng
pr
e
di
c
ti
ons
f
or
C
A
D
us
in
g
S
V
M
.
S
V
M
ha
s
be
tt
e
r
p
e
r
f
or
m
a
nc
e
th
a
n
a
ll
te
s
t
e
d
a
lg
or
it
hm
s
,
s
uc
h
a
s
de
c
is
io
n
tr
e
e
,
k
-
N
N
,
a
nd
n
a
ï
ve
ba
ye
s
.
T
h
e
m
a
in
c
ha
ll
e
nge
of
th
i
s
r
e
s
e
a
r
c
h
is
to
de
a
l
w
it
h
unba
l
a
nc
e
d
da
ta
a
nd
th
e
m
a
ny
f
e
a
tu
r
e
s
in
th
e
d
a
ta
s
e
t.
A
lt
hough
th
e
f
ir
s
t
te
s
t
e
r
s
w
it
hout
us
in
g
f
e
a
tu
r
e
s
e
le
c
ti
on
a
nd
m
ode
l
r
e
s
a
m
pl
in
g
f
e
a
tu
r
e
s
ha
ve
w
or
ke
d
w
e
ll
f
or
m
ode
li
ng
th
e
da
ta
,
th
e
r
e
is
s
uf
f
ic
ie
nt
di
s
ta
nc
e
be
twe
e
n t
he
e
xa
m
in
e
r
s
a
nd t
he
t
r
a
in
in
g a
c
c
ur
a
c
y.
I
de
a
ll
y, t
r
a
in
in
g a
nd
te
s
ti
ng a
c
c
ur
a
c
y a
r
e
r
e
la
ti
ve
ly
c
lo
s
e
.
W
e
us
e
e
xt
r
a
tr
e
e
c
la
s
s
-
ba
s
e
d
f
e
a
tu
r
e
s
e
le
c
ti
on
to
ge
ne
r
a
te
16
upda
te
d
f
e
a
tu
r
e
s
on
th
e
ta
r
ge
t
v
a
r
ia
bl
e
.
O
ur
te
s
ts
us
in
g
r
a
ndom
tr
e
e
s
c
la
s
s
if
ie
r
f
or
f
e
a
tu
r
e
s
e
le
c
ti
on a
nd
S
M
O
T
E
f
or
f
e
a
tu
r
e
r
e
s
a
m
pl
in
g
s
how
th
a
t
th
e
be
s
t
m
ode
l
pe
r
f
or
m
a
nc
e
i
s
t
he
te
s
ti
ng
a
c
c
ur
a
c
y of
88%
.
R
E
F
E
R
E
N
C
E
S
[
1]
“
H
e
a
r
t
di
s
e
a
s
e
r
i
s
k
f
a
c
t
or
s
,”
T
e
x
as
H
e
a
r
t
I
ns
t
i
t
ut
e
.
ht
t
ps
:
/
/
w
w
w
.t
e
xa
s
he
a
r
t
.or
g/
he
a
r
t
-
he
a
l
t
h/
he
a
r
t
-
i
nf
o
r
m
a
t
i
on
-
c
e
nt
e
r
/
t
opi
c
s
/
he
a
r
t
-
di
s
e
a
s
e
-
r
i
s
k
-
f
a
c
t
or
s
(
a
c
c
e
s
s
e
d
M
a
y 26, 2021)
.
[
2]
F
.
S
a
nc
hi
s
-
G
om
a
r
,
C
.
P
e
r
e
z
-
Q
ui
l
i
s
,
R
.
L
e
i
s
c
hi
k,
a
nd
A
.
L
uc
i
a
,
“
E
pi
de
m
i
ol
o
gy
of
c
or
ona
r
y
he
a
r
t
d
i
s
e
a
s
e
a
nd
a
c
ut
e
c
or
ona
r
y
s
yndr
om
e
,”
A
nnal
s
of
T
r
ans
l
at
i
onal
M
e
di
c
i
ne
, vol
. 4, no. 13, pp. 256
–
256, J
ul
. 2016, doi
:
10.21037/
a
t
m
.2016.06.33.
[
3]
“
C
or
ona
r
y a
r
t
e
r
y di
s
e
a
s
e
(
C
A
D
)
.”
ht
t
ps
:
/
/
w
w
w
.
c
dc
.gov/
he
a
r
t
di
s
e
a
s
e
/
c
or
ona
r
y_
a
d.ht
m
(
a
c
c
e
s
s
e
d M
a
y 26, 2021)
.
[
4]
H
.
A
ni
m
e
s
h,
K
.
M
.
S
ubr
a
t
a
,
G
.
A
m
i
t
,
M
.
A
r
kom
i
t
a
,
a
nd
A
.
M
ukhe
r
j
e
,
“
H
e
a
r
t
di
s
e
a
s
e
di
a
gnos
i
s
a
nd
pr
e
di
c
t
i
on
us
i
ng
m
a
c
hi
ne
l
e
a
r
ni
ng
a
nd
da
t
a
m
i
ni
ng
t
e
c
hni
que
s
:
a
r
e
vi
e
w
,”
A
dv
anc
e
s
i
n
C
om
put
at
i
onal
S
c
i
e
nc
e
s
and
T
e
c
hnol
ogy
,
vol
.
10,
no.
7,
pp.
2137
–
2159, 2017.
[
5]
R
.
M
a
he
s
hw
a
r
i
,
K
.
M
oudgi
l
,
H
.
P
a
r
e
kh,
a
nd
R
.
S
a
w
a
nt
,
“
A
m
a
c
hi
ne
l
e
a
r
ni
ng
ba
s
e
d
m
e
di
c
a
l
da
t
a
a
n
a
l
yt
i
c
s
a
nd
vi
s
ua
l
i
z
a
t
i
on
r
e
s
e
a
r
c
h
pl
a
t
f
or
m
,”
i
n
2018
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
C
ur
r
e
nt
T
r
e
nds
t
o
w
ar
ds
C
on
v
e
r
gi
ng
T
e
c
hnol
ogi
e
s
(
I
C
C
T
C
T
)
,
M
a
r
.
2018, pp. 1
–
5, doi
:
10.1109/
I
C
C
T
C
T
.2018.8550953
.
[
6]
L
.
M
uf
l
i
kha
h,
N
.
W
i
dodo,
W
.
F
.
M
a
hm
udy,
a
nd
S
ol
i
m
un,
“
P
r
e
di
c
t
i
on
of
l
i
ve
r
c
a
nc
e
r
ba
s
e
d
on
D
N
A
s
e
que
nc
e
u
s
i
ng
e
ns
e
m
bl
e
m
e
t
hod,”
i
n
2020
3r
d
I
nt
e
r
nat
i
onal
Se
m
i
nar
on
R
e
s
e
ar
c
h
of
I
nf
or
m
at
i
on
T
e
c
hnol
ogy
and
I
nt
e
l
l
i
ge
nt
Sy
s
t
e
m
s
(
I
SR
I
T
I
)
,
D
e
c
.
2020,
pp. 37
–
41, doi
:
10.1109/
I
S
R
I
T
I
51436.2020.9315341.
[
7]
A
.
R
i
dok,
N
.
W
i
dodo,
W
.
F
.
M
a
hm
udy,
a
nd
M
.
R
i
f
a
’
i
,
“
A
hybr
i
d
f
e
a
t
ur
e
s
e
l
e
c
t
i
on
on
A
I
R
S
m
e
t
hod
f
or
i
de
nt
i
f
yi
ng
br
e
a
s
t
c
a
nc
e
r
di
s
e
a
s
e
s
,”
I
nt
e
r
nat
i
onal
J
ou
r
nal
of
E
l
e
c
t
r
i
c
al
and
C
om
put
e
r
E
ngi
ne
e
r
i
ng
(
I
J
E
C
E
)
,
vol
.
11,
no.
1,
pp.
728
–
735,
F
e
b.
2021,
doi
:
10.11591/
i
j
e
c
e
.v11i
1.pp728
-
735.
[
8]
S
.
S
um
i
a
t
i
,
H
.
S
a
r
a
gi
h,
T
.
A
bdul
R
a
hm
a
n,
a
nd
A
.
T
r
i
a
yudi
,
“
E
xpe
r
t
s
ys
t
e
m
f
or
he
a
r
t
di
s
e
a
s
e
ba
s
e
d
on
e
l
e
c
t
r
oc
a
r
di
ogr
a
m
da
t
a
us
i
ng
c
e
r
t
a
i
nt
y
f
a
c
t
or
w
i
t
h
m
ul
t
i
pl
e
r
ul
e
,”
I
A
E
S
I
n
t
e
r
n
at
i
onal
J
our
nal
of
A
r
t
i
f
i
c
i
al
I
nt
e
l
l
i
ge
nc
e
(
I
J
-
A
I
)
,
vol
.
10,
no.
1,
pp.
43
–
50,
M
a
r
. 2021, doi
:
10.11591/
i
j
a
i
.v10.i
1.pp43
-
50.
[
9]
S
. S
a
h, “
M
a
c
hi
ne
l
e
a
r
ni
ng:
a
r
e
vi
e
w
of
l
e
a
r
ni
ng t
ype
s
,”
P
r
e
p
r
i
nt
s
, J
ul
. 2020, doi
:
10.20944/
pr
e
pr
i
nt
s
202007.0230.v1.
[
10]
W
.
Y
u
,
T
.
L
i
u,
R
.
V
a
l
de
z
,
M
.
G
w
i
nn,
a
nd
M
.
J
.
K
hour
y,
“
A
ppl
i
c
a
t
i
on
of
s
uppor
t
ve
c
t
or
m
a
c
hi
ne
m
ode
l
i
ng
f
or
pr
e
di
c
t
i
on
of
c
om
m
on
di
s
e
a
s
e
s
:
t
he
c
a
s
e
of
di
a
be
t
e
s
a
nd
pr
e
-
di
a
be
t
e
s
,
”
B
M
C
M
e
di
c
al
I
nf
or
m
at
i
c
s
and
D
e
c
i
s
i
on
M
ak
i
ng
,
vol
.
10,
no.
1,
D
e
c
.
2010, doi
:
10.1186
/
1472
-
6947
-
10
-
16.
[
11]
S
.
U
ddi
n,
A
.
K
ha
n,
M
.
E
.
H
os
s
a
i
n,
a
nd
M
.
A
.
M
oni
,
“
C
om
pa
r
i
ng
di
f
f
e
r
e
nt
s
upe
r
vi
s
e
d
m
a
c
hi
ne
l
e
a
r
ni
ng
a
l
gor
i
t
hm
s
f
or
di
s
e
a
s
e
pr
e
di
c
t
i
on,”
B
M
C
M
e
di
c
al
I
nf
or
m
at
i
c
s
and D
e
c
i
s
i
on M
ak
i
ng
, vol
. 19, no. 1, D
e
c
. 2019, doi
:
10.1186/
s
12911
-
019
-
1004
-
8.
[
12]
A
.
R
i
dok,
N
.
W
i
dodo,
W
.
F
.
M
a
hm
udy,
a
nd
M
.
R
i
f
a
i
,
“
F
C
-
S
V
M
:
D
N
A
bi
nd
i
ng
P
r
ot
e
i
ns
pr
e
di
c
t
i
on
w
i
t
h
a
ve
r
a
ge
bl
oc
ks
(
A
B
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
11
, N
o.
1
,
M
a
r
c
h
20
22
:
345
-
355
354
de
s
c
r
i
pt
or
s
us
i
ng
S
V
M
w
i
t
h F
C
f
e
a
t
ur
e
s
e
l
e
c
t
i
on,”
i
n
2019
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on Sus
t
ai
nabl
e
I
nf
or
m
at
i
on E
ngi
ne
e
r
i
ng
and
T
e
c
hnol
ogy
(
SI
E
T
)
,
S
e
p. 20
19, pp. 22
–
27, doi
:
10.1109/
S
I
E
T
48054.2019.8986
070.
[
13]
R
.
A
l
i
z
a
de
hs
a
ni
e
t
al
.
,
“
N
on
-
i
nva
s
i
ve
de
t
e
c
t
i
on
of
c
or
ona
r
y
a
r
t
e
r
y
di
s
e
a
s
e
i
n
hi
gh
-
r
i
s
k
pa
t
i
e
nt
s
ba
s
e
d
on
t
he
s
t
e
nos
i
s
pr
e
di
c
t
i
on
of
s
e
pa
r
a
t
e
c
or
ona
r
y
a
r
t
e
r
i
e
s
,”
C
om
put
e
r
M
e
t
hods
and
P
r
ogr
am
s
i
n
B
i
om
e
di
c
i
ne
,
vol
.
162,
pp.
119
–
127,
A
ug.
2018,
doi
:
10.1016/
j
.c
m
pb.2018.05.009.
[
14]
R
.
A
l
i
z
a
de
h
s
a
ni
e
t
al
.
,
“
D
i
a
gno
s
i
ng
c
or
ona
r
y
a
r
t
e
r
y
di
s
e
a
s
e
vi
a
d
a
t
a
m
i
ni
ng
a
l
gor
i
t
hm
s
by
c
on
s
i
de
r
i
ng
l
a
bor
a
t
or
y
a
nd
e
c
hoc
a
r
di
ogr
a
phy f
e
a
t
ur
e
s
,”
R
e
s
e
a
r
c
h i
n C
ar
di
o
v
as
c
ul
a
r
M
e
d
i
c
i
ne
, vol
. 2, no. 3, 2013, doi
:
10.5812/
c
a
r
di
ova
s
c
m
e
d.10888.
[
15]
R
.
A
l
i
z
a
de
h
s
a
ni
e
t
al
.
,
“
A
da
t
a
m
i
ni
ng
a
ppr
oa
c
h
f
or
di
a
gnos
i
s
of
c
or
ona
r
y
a
r
t
e
r
y
di
s
e
a
s
e
,”
C
om
put
e
r
M
e
t
hods
and
P
r
og
r
am
s
i
n
B
i
om
e
di
c
i
ne
, vol
. 111, no. 1, pp. 52
–
61, J
ul
. 2013, doi
:
10.1016/
j
.c
m
pb.2013.0
3.004.
[
16]
J
.
H
.
J
ol
ouda
r
i
e
t
al
.
,
“
C
or
ona
r
y
a
r
t
e
r
y
di
s
e
a
s
e
di
a
gnos
i
s
;
r
a
nki
ng
t
he
s
i
gni
f
i
c
a
nt
f
e
a
t
ur
e
s
us
i
ng
a
r
a
ndom
t
r
e
e
s
m
ode
l
,”
I
nt
e
r
nat
i
onal
J
our
nal
of
E
nv
i
r
onm
e
nt
al
R
e
s
e
ar
c
h and P
ubl
i
c
H
e
al
t
h
, vol
. 17, no. 3, J
a
n. 2020, doi
:
10
.3390/
i
j
e
r
ph17030731.
[
17]
S
a
vi
t
a
,
G
.
S
ha
r
m
a
,
G
.
R
a
ni
,
a
nd
V
.
S
.
D
ha
ka
,
“
E
f
f
i
c
i
e
nt
pr
e
di
c
t
i
ve
m
ode
l
l
i
ng
f
or
c
l
a
s
s
i
f
i
c
a
t
i
on
of
c
or
ona
r
y
a
r
t
e
r
y
d
i
s
e
a
s
e
s
u
s
i
ng
m
a
c
hi
ne
l
e
a
r
ni
ng
a
ppr
oa
c
h,”
I
O
P
C
onf
e
r
e
nc
e
Se
r
i
e
s
:
M
at
e
r
i
al
s
Sc
i
e
nc
e
and
E
ngi
ne
e
r
i
ng
,
vol
.
1099,
no
.
1,
p.
012068,
M
a
r
.
2021,
doi
:
10.1088/
1757
-
899X
/
1099/
1/
012068.
[
18]
K
.
E
.
H
e
i
ke
s
,
D
.
M
.
E
ddy,
B
.
A
r
onde
ka
r
,
a
nd
L
.
S
c
hl
e
s
s
i
nge
r
,
“
D
i
a
be
t
e
s
r
i
s
k
c
a
l
c
ul
a
t
or
:
a
s
i
m
pl
e
t
ool
f
or
de
t
e
c
t
i
ng
undi
a
gnos
e
d
di
a
be
t
e
s
a
nd pr
e
-
di
a
be
t
e
s
,
”
D
i
abe
t
e
s
C
ar
e
, vol
. 31, no. 5,
pp. 1040
–
1045,
M
a
y
2008, doi
:
10.2337/
dc
07
-
1150.
[
19]
K
.
S
ha
nka
r
,
S
.
K
.
L
a
ks
hm
a
na
pr
a
bu,
D
.
G
upt
a
,
A
.
M
a
s
e
l
e
no,
a
nd
V
.
H
.
C
.
de
A
l
buque
r
que
,
“
O
pt
i
m
a
l
f
e
a
t
ur
e
-
ba
s
e
d
m
ul
t
i
-
ke
r
ne
l
S
V
M
a
ppr
oa
c
h
f
or
t
hyr
oi
d
di
s
e
a
s
e
c
l
a
s
s
i
f
i
c
a
t
i
on,”
T
he
J
ou
r
nal
of
Supe
r
c
om
put
i
ng
,
vol
.
76,
no.
2,
pp.
1128
–
1143,
F
e
b.
2020,
doi
:
10.1007/
s
11227
-
018
-
2469
-
4.
[
20]
C
.
A
r
oe
f
,
Y
.
R
i
va
n,
a
nd
Z
.
R
u
s
t
a
m
,
“
C
om
pa
r
i
ng
r
a
ndom
f
or
e
s
t
a
nd
s
uppor
t
ve
c
t
or
m
a
c
hi
ne
s
f
or
br
e
a
s
t
c
a
nc
e
r
c
l
a
s
s
i
f
i
c
a
t
i
on,”
T
E
L
K
O
M
N
I
K
A
(
T
e
l
e
c
om
m
uni
c
at
i
on
C
om
put
i
ng
E
l
e
c
t
r
oni
c
s
and
C
o
nt
r
ol
)
,
vol
.
18,
no.
2,
A
pr
.
2020,
doi
:
10.12928/
t
e
l
kom
ni
ka
.v18i
2.14785.
[
21]
K
. D
a
va
gdor
j
, J
. S
. L
e
e
, V
.
H
. P
ha
m
, a
nd
K
. H
. R
yu,
“
A
c
om
pa
r
a
t
i
ve
a
na
l
ys
i
s
of
m
a
c
hi
ne
l
e
a
r
ni
ng m
e
t
hods
f
or
c
l
a
s
s
i
m
ba
l
a
nc
e
i
n
a
s
m
oki
ng c
e
s
s
a
t
i
on i
nt
e
r
ve
nt
i
on,”
A
ppl
i
e
d Sc
i
e
nc
e
s
, vol
. 10, no. 9,
M
a
y 2020,
doi
:
10.3390/
a
pp10093307.
[
22]
N
.
S
a
nt
os
o,
W
.
W
i
bow
o,
a
nd
H
.
H
i
km
a
w
a
t
i
,
“
I
nt
e
gr
a
t
i
on
of
s
ynt
he
t
i
c
m
i
nor
i
t
y
ove
r
s
a
m
pl
i
ng
t
e
c
hni
que
f
o
r
i
m
ba
l
a
nc
e
d
c
l
a
s
s
,”
I
ndone
s
i
an
J
our
nal
of
E
l
e
c
t
r
i
c
al
E
ngi
ne
e
r
i
ng
and
C
om
put
e
r
Sc
i
e
n
c
e
,
vo
l
.
13,
no.
1,
pp.
102
–
108,
J
a
n.
2019,
doi
:
10.11591/
i
j
e
e
c
s
.v13.i
1.pp102
-
108.
[
23]
K
.
D
a
va
gdor
j
,
J
.
S
.
L
e
e
,
K
.
H
.
P
a
r
k,
P
.
V
.
H
uy,
a
nd
K
.
H
.
R
yu,
“
S
ynt
he
t
i
c
o
ve
r
s
a
m
pl
i
ng
ba
s
e
d
de
c
i
s
i
on
s
uppor
t
f
r
a
m
e
w
or
k
t
o
s
ol
ve
c
l
a
s
s
i
m
ba
l
a
nc
e
pr
obl
e
m
i
n
s
m
oki
ng
c
e
s
s
a
t
i
on
pr
ogr
a
m
,”
I
nt
e
r
nat
i
onal
J
our
nal
of
A
ppl
i
e
d
Sc
i
e
nc
e
and
E
ngi
ne
e
r
i
ng
,
vol
.
17, no. 3, pp. 223
–
235, 2020.
[
24]
I
.
D
.
A
pos
t
ol
opoul
os
,
“
I
nve
s
t
i
ga
t
i
ng
t
h
e
s
ynt
h
e
t
i
c
m
i
nor
i
t
y
c
l
a
s
s
ove
r
s
a
m
pl
i
ng
t
e
c
hni
que
(
S
M
O
T
E
)
on
a
n
i
m
ba
l
a
nc
e
d
c
a
r
di
ova
s
c
ul
a
r
di
s
e
a
s
e
(
C
V
D
)
da
t
a
s
e
t
,”
I
nt
e
r
nat
i
onal
J
our
nal
of
E
ngi
ne
e
r
i
ng
A
ppl
i
e
d
Sc
i
e
nc
e
s
and
T
e
c
hnol
ogy
,
vol
.
4,
no.
9,
pp
.
431
–
434, J
a
n. 2020, doi
:
10.33564/
I
J
E
A
S
T
.2020.v04i
09.058.
[
25]
C
.
T
a
l
e
b,
M
.
K
ha
c
ha
b,
C
.
M
okb
e
l
,
a
nd
L
.
L
i
kf
or
m
a
n
-
S
ul
e
m
,
“
A
r
e
l
i
a
bl
e
m
e
t
hod
t
o
pr
e
di
c
t
pa
r
ki
ns
on’
s
di
s
e
a
s
e
s
t
a
g
e
a
n
d
pr
ogr
e
s
s
i
on
ba
s
e
d
on
h
a
ndw
r
i
t
i
ng
a
nd
r
e
-
s
a
m
pl
i
ng
a
ppr
oa
c
he
s
,
”
i
n
2018
I
E
E
E
2nd
I
nt
e
r
nat
i
onal
W
or
k
s
hop
on
A
r
abi
c
an
d
D
e
r
i
v
e
d Sc
r
i
pt
A
nal
y
s
i
s
and R
e
c
ogni
t
i
on (
A
SA
R
)
, M
a
r
. 2018, pp. 7
–
12, doi
:
10.
1109/
A
S
A
R
.2018.8480209.
[
26]
A
.
R
i
dok,
W
.
F
.
M
a
hm
udy,
a
nd
M
.
R
i
f
a
i
,
“
A
n
i
m
pr
ove
d
a
r
t
i
f
i
c
i
a
l
i
m
m
une
r
e
c
ogni
t
i
on
s
ys
t
e
m
w
i
t
h
f
a
s
t
c
or
r
e
l
a
t
i
on
ba
s
e
d
f
i
l
t
e
r
(
F
C
B
F
)
f
or
f
e
a
t
ur
e
s
e
l
e
c
t
i
on,”
i
n
2017 F
our
t
h
I
nt
e
r
nat
i
onal
C
onf
e
r
e
n
c
e
on
I
m
age
I
nf
or
m
at
i
on P
r
oc
e
s
s
i
ng
(
I
C
I
I
P
)
, D
e
c
. 2017, pp
.
1
–
6, doi
:
10.1109/
I
C
I
I
P
.2017.8313761.
[
27]
A
.
M
.
A
.
a
nd
P
.
A
.
T
hom
a
s
,
“
C
om
pa
r
a
t
i
ve
r
e
vi
e
w
o
f
f
e
a
t
ur
e
s
e
l
e
c
t
i
on
a
nd
c
l
a
s
s
i
f
i
c
a
t
i
on
m
ode
l
i
ng,”
i
n
2019
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
A
dv
anc
e
s
i
n
C
om
put
i
ng,
C
om
m
uni
c
at
i
on
and
C
o
nt
r
ol
(
I
C
A
C
3)
,
D
e
c
.
2019,
pp.
1
–
9,
doi
:
10.1109/
I
C
A
C
34759
0.2019.9036816.
[
28]
N
.
V
.
C
ha
w
l
a
,
K
.
W
.
B
o
w
ye
r
,
L
.
O
.
H
a
l
l
,
a
nd
W
.
P
.
K
e
ge
l
m
e
ye
r
,
“
S
M
O
T
E
:
s
ynt
he
t
i
c
m
i
nor
i
t
y
ove
r
-
s
a
m
pl
i
ng
t
e
c
hni
que
,”
J
our
nal
of
A
r
t
i
f
i
c
i
al
I
nt
e
l
l
i
ge
nc
e
R
e
s
e
ar
c
h
, vol
. 16, pp. 321
–
357, J
un. 2002, doi
:
10.1613/
j
a
i
r
.953.
[
29]
H
.
H
e
, Y
.
B
a
i
,
E
. A
.
G
a
r
c
i
a
,
a
nd S
.
L
i
, “
A
D
A
S
Y
N
:
a
da
pt
i
ve
s
ynt
he
t
i
c
s
a
m
pl
i
n
g a
ppr
oa
c
h f
or
i
m
ba
l
a
nc
e
d l
e
a
r
ni
ng,”
i
n
2008 I
E
E
E
I
nt
e
r
nat
i
onal
J
oi
nt
C
onf
e
r
e
nc
e
on
N
e
ur
al
N
e
t
w
or
k
s
(
I
E
E
E
W
or
l
d
C
ongr
e
s
s
on
C
om
put
at
i
onal
I
nt
e
l
l
i
ge
nc
e
)
,
J
un.
2008,
pp.
1322
–
1328, doi
:
10.1109/
I
J
C
N
N
.2008.4633969.
[
30]
I
.
C
.
D
i
pt
o,
T
.
I
s
l
a
m
,
H
.
M
.
M
.
R
a
hm
a
n,
a
nd
M
.
A
.
R
a
hm
a
n,
“
C
om
p
a
r
i
s
on
of
di
f
f
e
r
e
nt
m
a
c
hi
ne
l
e
a
r
ni
ng
a
l
gor
i
t
h
m
s
f
or
t
he
pr
e
di
c
t
i
on
of
c
or
ona
r
y
a
r
t
e
r
y
di
s
e
a
s
e
,”
J
ou
r
nal
of
D
at
a
A
nal
y
s
i
s
and
I
nf
or
m
at
i
on
P
r
oc
e
s
s
i
ng
,
vol
.
8,
no.
2,
pp.
41
–
68,
2020,
doi
:
10.4236/
j
da
i
p.2020.82003.
[
31]
G
.
A
.
F
a
ns
hur
i
A
l
f
a
r
i
s
y,
W
.
F
.
M
a
hm
udy,
a
nd
M
.
H
.
N
a
t
s
i
r
,
“
G
ood
pa
r
a
m
e
t
e
r
s
f
or
P
S
O
i
n
opt
i
m
i
z
i
ng
l
a
yi
ng
he
n
di
e
t
,”
I
nt
e
r
nat
i
onal
J
our
nal
of
E
l
e
c
t
r
i
c
al
and
C
om
put
e
r
E
ngi
ne
e
r
i
ng
(
I
J
E
C
E
)
,
vol
.
8,
no.
4,
pp.
2419
–
2432,
A
ug.
2018,
doi
:
10.11591/
i
j
e
c
e
.v8i
4.pp2419
-
2432.
[
32]
I
.
S
ya
r
i
f
,
A
.
P
r
uge
l
-
B
e
nne
t
t
,
a
nd
G
.
W
i
l
l
s
,
“
S
V
M
pa
r
a
m
e
t
e
r
opt
i
m
i
z
a
t
i
on
us
i
ng
gr
i
d
s
e
a
r
c
h
a
nd
ge
ne
t
i
c
a
l
gor
i
t
hm
t
o
i
m
pr
ove
c
l
a
s
s
i
f
i
c
a
t
i
on
pe
r
f
or
m
a
nc
e
,”
T
E
L
K
O
M
N
I
K
A
(
T
e
l
e
c
om
m
uni
c
at
i
on
C
om
put
i
ng
E
l
e
c
t
r
oni
c
s
and
C
ont
r
ol
)
,
vol
.
14,
no.
4,
pp.
1502
–
1509, D
e
c
. 2016, doi
:
10.12928/
t
e
l
kom
ni
ka
.v14i
4.3956.
[
33]
L
.
M
uf
l
i
kha
h
a
nd
D
.
J
.
H
a
r
ya
nt
o,
“
H
i
gh
pe
r
f
or
m
a
nc
e
of
pol
ynom
i
a
l
ke
r
ne
l
a
t
S
V
M
a
l
gor
i
t
hm
f
or
s
e
nt
i
m
e
nt
a
na
l
ys
i
s
,”
J
our
nal
o
f
I
nf
or
m
at
i
on T
e
c
hnol
o
gy
and C
om
put
e
r
Sc
i
e
n
c
e
, vol
. 3, no. 2, pp. 194
–
201, N
o
v. 2018, doi
:
10.25126/
j
i
t
e
c
s
.20183260.
B
I
O
G
R
A
P
H
I
E
S
O
F
A
U
T
H
O
R
S
Kuncahyo
Setyo
Nugroho
received
a
bachelor
of
computer
d
egree
from
the
Department
of
Informatics
Engineering,
Faculty
of
Engineering,
Widyagama
University,
Indonesia, i
n 2019. He i
s
currently pu
rsuing a
master'
s
degree at the
D
epartment of In
formatics
Engineering,
Faculty
of
Co
mputer
Science,
Brawijaya
University,
In
donesia.
He
is
a
member
of
the
Intelligent
Systems
Research
Laboratory,
with
interest
in
affec
tive
computing.
He
also
has
research
interests
ar
e
in
machine
learning,
deep
learning,
and
natu
ral
language
processing.
He
can be cont
acted at em
ail:
ksnugroho26@
gmail.com or ksnugroho@
student.ub.ac.id.
Evaluation Warning : The document was created with Spire.PDF for Python.