I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
15
,
No
.
6
,
Decem
b
er
20
25
,
p
p
.
5
5
5
5
~
5
5
6
9
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
6
.
pp
5
5
5
5
-
5
5
6
9
5555
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
A
hy
brid
DM
O
-
C
NN
-
LST
M
frame
wo
rk f
o
r
feature
selection
a
nd dia
betes
pred
iction: a d
eep l
ea
rning
perspec
tive
M
uta
s
em
K
.
Als
m
a
di
1
,
G
ha
it
h M
.
J
a
ra
da
t
2
,
T
a
riq A
ls
a
lla
k
2
,
M
a
lek A
lza
qeba
h
3
,4
,
Sa
na
J
a
wa
rneh
5
,
H
a
y
a
t
Alf
a
g
ha
m
1
,
J
eha
d Alqurni
6
,
Usa
m
a
A.
B
a
da
wi
1
,
L
a
t
if
a
Abdu
lla
h Alm
us
f
a
r
7
1
D
e
p
a
r
t
me
n
t
o
f
M
I
S
,
C
o
l
l
e
g
e
o
f
A
p
p
l
i
e
d
S
t
u
d
i
e
s a
n
d
C
o
mm
u
n
i
t
y
S
e
r
v
i
c
e
,
I
mam
A
b
d
u
l
r
a
h
m
a
n
B
i
n
F
a
i
sa
l
U
n
i
v
e
r
si
t
y
,
D
a
mm
a
m
,
S
a
u
d
i
A
r
a
b
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
I
n
f
o
r
ma
t
i
o
n
S
y
st
e
ms,
C
o
l
l
e
g
e
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
s
a
n
d
I
n
f
o
r
mat
i
c
s
,
A
m
man
A
r
a
b
U
n
i
v
e
r
si
t
y
,
M
u
b
i
s
-
A
mm
a
n
,
J
o
r
d
a
n
3
D
e
p
a
r
t
me
n
t
o
f
M
a
t
h
e
m
a
t
i
c
s
,
C
o
l
l
e
g
e
o
f
S
c
i
e
n
c
e
,
I
mam
A
b
d
u
l
r
a
h
ma
n
B
i
n
F
a
i
sa
l
U
n
i
v
e
r
si
t
y
,
D
a
mm
a
m
,
S
a
u
d
i
A
r
a
b
i
a
4
B
a
s
i
c
a
n
d
A
p
p
l
i
e
d
S
c
i
e
n
t
i
f
i
c
R
e
se
a
r
c
h
C
e
n
t
e
r
,
I
mam
A
b
d
u
l
r
a
h
m
a
n
B
i
n
F
a
i
sal
U
n
i
v
e
r
s
i
t
y
,
D
a
mm
a
m,
S
a
u
d
i
A
r
a
b
i
a
5
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
D
e
p
a
r
t
me
n
t
,
A
p
p
l
i
e
d
C
o
l
l
e
g
e
,
I
mam
A
b
d
u
l
r
a
h
m
a
n
B
i
n
F
a
i
sa
l
U
n
i
v
e
r
si
t
y
,
D
a
mm
a
m
,
S
a
u
d
i
A
r
a
b
i
a
6
D
e
p
a
r
t
me
n
t
o
f
E
d
u
c
a
t
i
o
n
a
l
Te
c
h
n
o
l
o
g
i
e
s,
C
o
l
l
e
g
e
o
f
E
d
u
c
a
t
i
o
n
,
I
mam
A
b
d
u
l
r
a
h
ma
n
B
i
n
F
a
i
sa
l
U
n
i
v
e
r
s
i
t
y
,
D
a
mm
a
m,
S
a
u
d
i
A
r
a
b
i
a
7
D
e
p
a
r
t
m
e
n
t
o
f
M
a
n
a
g
e
m
e
n
t
I
n
f
o
r
m
a
t
i
o
n
S
y
s
t
e
ms
,
C
o
l
l
e
g
e
o
f
B
u
s
i
n
e
s
s
A
d
mi
n
i
s
t
r
a
t
i
o
n
,
I
m
a
m
A
b
d
u
l
r
a
h
m
a
n
B
i
n
F
a
i
sa
l
U
n
i
v
e
r
s
i
t
y
(
I
A
U
)
,
D
a
mm
a
n
,
S
a
u
d
i
A
r
a
b
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
3
,
2
0
2
4
R
ev
is
ed
Sep
6
,
2
0
2
5
Acc
ep
ted
Sep
1
4
,
2
0
2
5
T
h
e
e
a
r
l
y
a
n
d
a
c
c
u
r
a
t
e
p
r
e
d
i
c
t
i
o
n
o
f
d
i
a
b
e
t
e
s
m
e
l
l
i
t
u
s
r
e
m
a
i
n
s
a
s
i
g
n
i
f
i
c
a
n
t
c
h
a
l
l
e
n
g
e
i
n
c
l
i
n
i
c
a
l
d
e
c
i
s
i
o
n
-
m
a
k
i
n
g
d
u
e
t
o
t
h
e
h
i
g
h
d
i
m
e
n
s
i
o
n
a
l
i
t
y
,
n
o
i
s
e
,
a
n
d
h
e
t
e
r
o
g
e
n
e
i
t
y
o
f
m
e
d
i
c
a
l
d
a
t
a
.
T
h
i
s
s
t
u
d
y
p
r
o
p
o
s
e
s
a
n
o
v
e
l
h
y
b
r
i
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
f
r
a
m
e
w
o
r
k
t
h
a
t
i
n
t
e
g
r
a
t
e
s
t
h
e
d
w
a
r
f
m
o
n
g
o
o
s
e
o
p
t
i
m
i
z
a
t
i
o
n
(
D
M
O
)
a
l
g
o
r
i
t
h
m
f
o
r
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
w
i
t
h
a
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
–
l
o
n
g
s
h
o
r
t
-
t
e
r
m
m
e
m
o
r
y
(
C
N
N
-
L
S
T
M
)
d
e
e
p
l
e
a
r
n
i
n
g
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
p
r
e
d
i
c
t
i
v
e
m
o
d
e
l
i
n
g
.
T
h
e
D
M
O
a
l
g
o
r
i
t
h
m
i
s
e
m
p
l
o
y
e
d
t
o
i
n
t
e
l
l
i
g
e
n
t
l
y
s
e
l
e
c
t
t
h
e
m
o
s
t
i
n
f
o
r
m
a
t
i
v
e
s
u
b
s
e
t
o
f
f
e
a
t
u
r
e
s
f
r
o
m
a
l
a
r
g
e
-
s
c
a
l
e
d
i
a
b
e
t
e
s
d
a
t
a
s
e
t
c
o
l
l
e
c
t
e
d
f
r
o
m
1
3
0
U
.
S
.
h
o
s
p
i
t
a
l
s
o
v
e
r
a
1
0
-
y
e
a
r
p
e
r
i
o
d
.
T
h
e
s
e
o
p
t
i
m
i
z
e
d
f
e
a
t
u
r
e
s
a
r
e
t
h
e
n
p
r
o
c
e
s
s
e
d
b
y
t
h
e
C
N
N
-
L
S
T
M
m
o
d
e
l
,
w
h
i
c
h
c
o
m
b
i
n
e
s
s
p
a
t
i
a
l
p
a
t
t
e
r
n
r
e
c
o
g
n
i
t
i
o
n
a
n
d
t
e
m
p
o
r
a
l
s
e
q
u
e
n
c
e
l
e
a
r
n
i
n
g
t
o
e
n
h
a
n
c
e
p
r
e
d
i
c
t
i
v
e
a
c
c
u
r
a
c
y
.
E
x
t
e
n
s
i
v
e
e
x
p
e
r
i
m
e
n
t
s
w
e
r
e
c
o
n
d
u
c
t
e
d
a
n
d
c
o
m
p
a
r
e
d
a
g
a
i
n
s
t
t
r
a
d
i
t
i
o
n
a
l
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
m
o
d
e
l
s
(
l
o
g
i
s
t
i
c
r
e
g
r
e
s
s
i
o
n
,
r
a
n
d
o
m
f
o
r
e
s
t
,
X
G
B
o
o
s
t
)
,
b
a
s
e
l
i
n
e
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
s
(
M
L
P
,
s
t
a
n
d
a
l
o
n
e
C
N
N
,
s
t
a
n
d
a
l
o
n
e
L
S
T
M
)
,
a
n
d
s
t
a
t
e
-
of
-
t
h
e
-
a
r
t
h
y
b
r
i
d
c
l
a
s
s
i
f
i
e
r
s
.
T
h
e
p
r
o
p
o
s
e
d
D
M
O
-
C
N
N
-
L
S
T
M
m
o
d
e
l
a
c
h
i
e
v
e
d
t
h
e
h
i
g
h
e
s
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
p
e
r
f
o
r
m
a
n
c
e
w
i
t
h
a
n
a
c
c
u
r
a
c
y
o
f
9
6
.
1
%
,
F
1
-
s
c
o
r
e
o
f
9
4
.
6
%
,
a
n
d
R
O
C
-
A
U
C
o
f
0
.
9
6
,
s
i
g
n
i
f
i
c
a
n
t
l
y
o
u
t
p
e
r
f
o
r
m
i
n
g
o
t
h
e
r
m
o
d
e
l
s
.
A
d
d
i
t
i
o
n
a
l
a
n
a
l
y
s
e
s
,
i
n
c
l
u
d
i
n
g
c
o
n
f
u
s
i
o
n
m
a
t
r
i
x
,
R
O
C
c
u
r
v
e
s
,
t
r
a
i
n
i
n
g
c
o
n
v
e
r
g
e
n
c
e
p
l
o
t
s
,
a
n
d
s
t
a
t
i
s
t
i
c
a
l
e
v
a
l
u
a
t
i
o
n
s
c
o
n
f
i
r
m
t
h
e
r
o
b
u
s
t
n
e
s
s
a
n
d
g
e
n
e
r
a
l
i
z
a
b
i
l
i
t
y
o
f
t
h
e
a
p
p
r
o
a
c
h
.
T
h
e
s
e
f
i
n
d
i
n
g
s
s
u
g
g
e
s
t
t
h
a
t
t
h
e
D
M
O
-
C
N
N
-
L
S
T
M
f
r
a
m
e
w
o
r
k
o
f
f
e
r
s
a
p
o
w
e
r
f
u
l
a
n
d
i
n
t
e
r
p
r
e
t
a
b
l
e
t
o
o
l
f
o
r
i
n
t
e
l
l
i
g
e
n
t
d
i
a
b
e
t
e
s
p
r
e
d
i
c
t
i
o
n
,
w
i
t
h
s
t
r
o
n
g
p
o
t
e
n
t
i
a
l
f
o
r
i
n
t
e
g
r
a
t
i
o
n
i
n
t
o
r
e
a
l
-
w
o
r
l
d
c
l
i
n
i
c
a
l
d
e
c
i
s
i
o
n
-
s
u
p
p
o
r
t
s
y
s
t
e
m
s
.
K
ey
w
o
r
d
s
:
C
lin
ical
d
ec
is
io
n
s
u
p
p
o
r
t
s
y
s
tem
C
NN
-
L
STM
Diab
etes p
r
ed
ictio
n
Dwa
r
f
m
o
n
g
o
o
s
e
o
p
tim
izatio
n
Featu
r
e
s
elec
tio
n
Hy
b
r
id
class
if
icatio
n
m
o
d
el
Me
d
ical
d
iag
n
o
s
is
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
:
Mu
tasem
K.
Als
m
ad
i
Dep
ar
tm
en
t o
f
MI
S,
C
o
lleg
e
o
f
Ap
p
lied
Stu
d
ies an
d
C
o
m
m
u
n
ity
Ser
v
ice,
I
m
a
m
Ab
d
u
lr
ah
m
an
B
in
Fais
al
Un
iv
er
s
ity
Dam
m
am
,
Sau
d
i A
r
ab
ia
E
m
ail: m
k
alsma
d
i@
iau
.
ed
u
.
s
a
1.
I
NT
RO
D
UCT
I
O
N
Diab
etes
m
ellitu
s
(
DM
)
is
a
p
er
v
asiv
e
an
d
ch
r
o
n
ic
d
is
ea
s
e
af
f
ec
tin
g
m
o
r
e
th
an
4
0
0
m
illi
o
n
p
eo
p
le
g
lo
b
ally
,
with
in
cr
ea
s
in
g
p
r
e
v
alen
ce
p
ar
ticu
lar
ly
in
d
ev
el
o
p
in
g
n
atio
n
s
[
1
]
.
E
f
f
ec
tiv
e
ea
r
ly
p
r
ed
ictio
n
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
6
,
Decem
b
e
r
20
25
:
5
5
5
5
-
5
5
6
9
5556
d
iab
etes
p
lay
s
a
v
ital
r
o
le
in
tim
ely
m
ed
ical
in
ter
v
en
ti
o
n
,
w
h
ich
s
ig
n
if
ican
tly
im
p
r
o
v
es
p
a
tien
t
o
u
tco
m
es
an
d
r
ed
u
ce
s
h
ea
lth
ca
r
e
c
o
s
ts
.
Ho
wev
er
,
th
is
task
is
o
f
ten
co
m
p
licated
b
y
h
i
g
h
-
d
i
m
en
s
io
n
al
m
ed
ical
d
ata,
n
o
is
e,
class
im
b
alan
ce
,
an
d
ir
r
elev
a
n
t f
ea
tu
r
e
s
th
at
ca
n
d
e
g
r
ad
e
m
o
d
el
p
er
f
o
r
m
an
ce
[
2
]
.
T
r
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
(
ML
)
m
o
d
els
s
u
ch
as
d
ec
is
i
o
n
tr
ee
s
(
DT
)
,
lo
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
,
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
es
(
SV
M)
,
an
d
m
u
ltil
ay
er
p
er
ce
p
t
r
o
n
n
eu
r
al
n
etwo
r
k
s
(
ML
P)
h
av
e
s
h
o
wn
p
r
o
m
is
e
in
d
iab
etes
class
if
icatio
n
[
3
]
.
Ho
wev
er
,
t
h
ese
m
o
d
els
o
f
ten
s
u
f
f
er
f
r
o
m
o
v
er
f
itti
n
g
,
p
o
o
r
g
en
er
aliza
tio
n
,
an
d
d
ep
en
d
e
n
ce
o
n
m
a
n
u
al
f
ea
tu
r
e
en
g
in
ee
r
in
g
.
Hy
b
r
i
d
m
eth
o
d
s
in
co
r
p
o
r
atin
g
o
p
tim
izatio
n
tech
n
iq
u
es
f
o
r
f
ea
tu
r
e
s
elec
tio
n
with
r
o
b
u
s
t c
lass
if
ier
s
h
av
e
r
ec
en
tly
g
ain
ed
atten
tio
n
.
T
h
is
s
tu
d
y
p
r
o
p
o
s
es
a
n
o
v
el
ap
p
r
o
ac
h
th
at
co
m
b
in
es
th
e
d
war
f
m
o
n
g
o
o
s
e
o
p
tim
izatio
n
(
DM
O)
alg
o
r
ith
m
[
4
]
f
o
r
au
t
o
m
atic
f
ea
tu
r
e
s
elec
tio
n
with
a
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
-
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
C
NN
-
L
STM
)
d
ee
p
lear
n
in
g
m
o
d
el
f
o
r
class
if
icatio
n
.
T
h
e
DM
O
alg
o
r
ith
m
,
in
s
p
ir
ed
b
y
t
h
e
h
u
n
tin
g
b
eh
av
io
r
o
f
d
war
f
m
o
n
g
o
o
s
es,
o
f
f
er
s
d
y
n
am
ic
ex
p
lo
r
atio
n
an
d
e
x
p
lo
it
atio
n
ca
p
ab
ilit
ies
f
o
r
id
en
tify
i
n
g
th
e
m
o
s
t
r
elev
an
t
f
ea
tu
r
e
s
u
b
s
ets
[
4
]
.
Me
a
n
wh
ile,
C
NN
-
L
STM
ar
ch
itectu
r
e
c
ap
tu
r
es
b
o
t
h
s
p
atial
an
d
tem
p
o
r
al
r
elatio
n
s
h
ip
s
i
n
p
atien
t d
ata,
im
p
r
o
v
in
g
p
r
e
d
ic
tio
n
ac
cu
r
ac
y
[
5
]
.
T
h
e
r
em
ain
d
e
r
o
f
th
is
p
ap
e
r
is
o
r
g
an
ized
as
f
o
llo
ws:
s
ec
tio
n
2
r
ev
iews
r
elev
an
t
liter
atu
r
e
o
n
d
iab
etes
p
r
ed
ictio
n
a
n
d
o
p
tim
izatio
n
alg
o
r
ith
m
s
.
Sectio
n
3
d
etails
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
.
Sectio
n
4
p
r
esen
ts
ex
p
er
im
en
tal
r
esu
lts
an
d
co
m
p
ar
ativ
e
an
al
y
s
is
.
Sectio
n
5
c
o
n
clu
d
es
with
k
ey
f
in
d
in
g
s
a
n
d
f
u
tu
r
e
r
esear
ch
d
ir
ec
t
io
n
s
.
2.
L
I
E
RA
T
UR
E
RE
V
I
E
W
2
.
1
.
Cla
s
s
ica
l
ma
chine le
a
rning
f
o
r
dia
bet
es pre
dict
io
n
T
r
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
(
ML
)
ap
p
r
o
ac
h
es,
s
u
ch
as
lo
g
is
tic
r
eg
r
ess
io
n
,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
(
SVM)
,
r
an
d
o
m
f
o
r
est,
an
d
g
r
ad
ien
t
b
o
o
s
tin
g
(
e.
g
.
,
XGBo
o
s
t)
,
h
av
e
b
ee
n
th
e
w
o
r
k
h
o
r
s
es
o
f
ea
r
ly
d
iab
etes
p
r
ed
ictio
n
r
esear
ch
,
y
ield
in
g
r
esu
lts
with
v
ar
y
in
g
d
eg
r
ee
s
o
f
s
u
cc
ess
.
T
h
eir
p
o
p
u
lar
it
y
s
tem
s
f
r
o
m
r
elativ
e
in
ter
p
r
etab
ilit
y
,
co
m
p
u
tatio
n
a
l
ef
f
icien
cy
,
a
n
d
s
tr
o
n
g
p
er
f
o
r
m
an
ce
o
n
s
m
aller
,
cu
r
ated
d
atasets
.
T
h
is
is
ex
em
p
lifie
d
b
y
s
tu
d
ies
lik
e
t
h
at
o
f
[
6
]
,
w
h
o
c
o
n
d
u
cted
c
o
m
p
ar
ativ
e
an
aly
s
es
o
f
m
u
ltip
le
class
if
ier
s
,
with
en
s
em
b
le
m
eth
o
d
s
lik
e
r
an
d
o
m
f
o
r
est
an
d
g
r
ad
ien
t
b
o
o
s
tin
g
r
ep
o
r
ted
ly
ac
h
iev
in
g
ac
c
u
r
ac
i
es
as
h
ig
h
as
9
8
.
8
%
o
n
s
p
ec
if
ic,
o
f
ten
p
r
e
-
p
r
o
ce
s
s
ed
d
atasets
.
Ho
wev
er
,
th
ese
ex
ce
p
tio
n
ally
h
ig
h
r
esu
lts
f
r
e
q
u
en
tly
m
ask
c
r
itical
lim
itatio
n
s
th
at
b
ec
o
m
e
ap
p
ar
en
t
u
n
d
er
r
ig
o
r
o
u
s
s
cr
u
tin
y
.
A
p
r
i
m
ar
y
is
s
u
e
is
th
e
p
r
o
p
en
s
ity
f
o
r
o
v
er
f
itti
n
g
,
wh
er
e
m
o
d
els ex
ce
l o
n
th
e
d
ata
th
e
y
wer
e
tr
ain
ed
o
n
b
u
t f
ail
to
m
ai
n
tain
p
er
f
o
r
m
a
n
ce
o
n
ex
ter
n
al
v
alid
atio
n
s
ets o
r
m
o
r
e
h
ete
r
o
g
en
eo
u
s
r
ea
l
-
wo
r
ld
d
ata.
T
h
is
lack
o
f
g
e
n
er
aliza
b
ilit
y
is
o
f
ten
c
o
m
p
o
u
n
d
e
d
b
y
a
d
ep
en
d
en
ce
o
n
m
an
u
al
f
ea
t
u
r
e
en
g
in
ee
r
in
g
an
d
th
e
ab
s
en
ce
o
f
r
o
b
u
s
t,
em
b
e
d
d
ed
f
ea
tu
r
e
s
elec
tio
n
m
ec
h
an
is
m
s
[
7
]
.
I
n
m
a
n
y
s
tu
d
ies,
f
ea
t
u
r
e
s
elec
tio
n
is
tr
ea
ted
as
a
s
ep
ar
ate
p
r
e
-
p
r
o
ce
s
s
in
g
s
t
ep
u
s
in
g
f
ilter
m
eth
o
d
s
(
e.
g
.
,
co
r
r
elatio
n
-
b
ased
)
o
r
is
h
a
n
d
l
ed
im
p
licitly
b
y
th
e
m
o
d
el
(
e.
g
.
,
f
ea
t
u
r
e
im
p
o
r
ta
n
ce
in
r
an
d
o
m
f
o
r
est
)
with
o
u
t
a
d
ed
icate
d
o
p
tim
izatio
n
p
r
o
c
ess
tai
lo
r
ed
to
th
e
m
o
d
el'
s
ar
ch
itectu
r
e.
T
h
is
ca
n
lead
to
th
e
in
clu
s
io
n
o
f
r
ed
u
n
d
a
n
t
o
r
n
o
is
y
f
ea
tu
r
es
t
h
at
d
eg
r
ad
e
m
o
d
el
p
er
f
o
r
m
an
ce
an
d
o
b
s
cu
r
e
th
e
m
o
s
t c
lin
ically
r
elev
an
t p
r
ed
ic
to
r
s
.
T
h
e
lim
itatio
n
s
o
f
th
ese
co
n
v
en
tio
n
al
m
eth
o
d
s
ar
e
f
u
r
t
h
er
h
i
g
h
lig
h
ted
b
y
m
o
r
e
r
ec
en
t
b
e
n
c
h
m
ar
k
in
g
s
tu
d
ies.
Fo
r
in
s
tan
ce
,
[
8
]
r
ep
o
r
ted
a
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
(
M
L
P)
ac
cu
r
ac
y
o
f
7
7
.
6
%,
wh
ile
[
9
]
ac
h
ie
v
ed
7
7
.
5
%
u
s
in
g
an
ML
P
o
n
th
e
class
ic
b
u
t
lim
ited
Pima
I
n
d
ian
Diab
etes
d
ataset.
T
h
ese
m
o
r
e
m
o
d
est
an
d
v
ar
iab
l
e
p
er
f
o
r
m
an
ce
m
etr
ics ar
e
a
r
g
u
a
b
ly
m
o
r
e
r
ef
lectiv
e
o
f
th
e
c
h
allen
g
es in
h
er
en
t i
n
clin
ical
d
ata
.
T
h
u
s
,
p
r
ev
io
u
s
s
tu
d
ies
u
n
d
er
s
co
r
e
a
p
r
ess
in
g
n
ee
d
to
m
o
v
e
b
ey
o
n
d
th
ese
co
n
v
en
tio
n
al
ap
p
r
o
ac
h
es.
T
h
e
co
r
e
s
h
o
r
tco
m
in
g
s
ar
e
t
h
r
ee
f
o
ld
:
i
n
ab
ilit
y
to
a
u
to
n
o
m
o
u
s
ly
lear
n
f
ea
tu
r
es;
s
tatic
m
o
d
elin
g
p
ar
ad
i
g
m
;
s
tr
u
g
g
le
with
h
ig
h
-
d
im
en
s
io
n
ality
.
C
o
n
s
eq
u
en
tly
,
th
ese
r
el
ativ
ely
m
o
d
est
r
esu
lts
an
d
in
h
er
en
t
lim
itatio
n
s
s
tr
o
n
g
ly
s
u
g
g
est
th
e
n
ec
ess
ity
f
o
r
m
o
r
e
s
o
p
h
is
ticated
,
au
t
o
m
ated
,
an
d
h
o
lis
tic
ap
p
r
o
ac
h
es.
T
h
er
e
is
a
clea
r
im
p
er
ativ
e
f
o
r
f
r
a
m
ewo
r
k
s
th
at
ca
n
in
tellig
en
tly
h
an
d
le
f
ea
tu
r
e
s
elec
tio
n
th
r
o
u
g
h
in
te
g
r
ated
o
p
ti
m
izatio
n
alg
o
r
ith
m
s
,
an
d
s
im
u
ltan
eo
u
s
ly
ca
p
tu
r
e
th
e
co
m
p
lex
s
p
atial
in
ter
ac
tio
n
s
an
d
tem
p
o
r
al
d
ep
en
d
en
cies
with
in
p
atien
t d
ata
to
ac
h
ie
v
e
r
o
b
u
s
t,
g
en
er
aliza
b
le,
a
n
d
clin
ically
a
ctio
n
ab
le
p
r
e
d
ictio
n
s
.
2
.
2
.
Dee
p
lea
rning
a
pp
ro
a
ch
es
Dee
p
lear
n
in
g
(
DL
)
m
et
h
o
d
s
h
av
e
em
e
r
g
ed
as
a
p
o
wer
f
u
l
p
ar
ad
i
g
m
to
o
v
er
c
o
m
e
th
e
in
h
er
en
t
lim
itatio
n
s
o
f
class
ical
m
ac
h
in
e
lear
n
in
g
,
p
r
im
ar
ily
b
y
au
to
m
atin
g
th
e
f
ea
tu
r
e
e
x
tr
ac
tio
n
p
r
o
ce
s
s
an
d
lear
n
in
g
co
m
p
lex
,
n
o
n
-
lin
ea
r
h
ier
ar
ch
ies
with
in
d
ata.
T
h
is
ca
p
ab
il
ity
is
ev
id
en
ce
d
b
y
s
tu
d
ies
s
u
ch
as
[
1
0
]
,
wh
o
en
g
in
ee
r
ed
an
im
p
r
o
v
e
d
ar
tif
icial
n
eu
r
al
n
etwo
r
k
(
ANN)
ar
ch
itectu
r
e
t
o
ac
h
ie
v
e
a
n
o
tab
le
9
3
%
ac
cu
r
ac
y
,
s
ig
n
if
ican
tly
o
u
tp
e
r
f
o
r
m
in
g
m
an
y
tr
ad
itio
n
al
m
o
d
els.
Fu
r
th
er
m
o
r
e,
t
h
e
ap
p
licatio
n
o
f
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NNs),
wh
ile
d
o
m
i
n
an
t
in
c
o
m
p
u
te
r
v
is
io
n
,
h
as
b
ee
n
cr
ea
tiv
ely
a
d
ap
ted
f
o
r
s
tr
u
ctu
r
ed
clin
ical
d
ata.
B
y
tr
ea
tin
g
a
p
atien
t'
s
f
ea
tu
r
e
v
ec
to
r
as
a
o
n
e
-
d
im
en
s
io
n
a
l
s
p
atial
m
ap
,
m
an
y
s
tu
d
ies
i
n
clu
d
in
g
[
1
1
]
h
av
e
d
em
o
n
s
tr
ated
th
at
C
NNs
ca
n
ef
f
ec
tiv
ely
id
en
tif
y
an
d
lear
n
in
tr
icate
,
lo
ca
l
s
p
atial
co
r
r
elatio
n
s
an
d
in
ter
ac
tio
n
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
DMO
-
CNN
-
LS
T
M fr
a
mewo
r
k
fo
r
fea
tu
r
e
s
elec
tio
n
a
n
d
d
i
a
b
etes
…
(
Mu
ta
s
em
K
.
A
ls
ma
d
i
)
5557
b
etwe
en
clin
ical
f
ea
tu
r
es.
Su
ch
as
th
e
r
elatio
n
s
h
ip
b
etwe
en
Hb
A1
c
lev
els,
ag
e,
an
d
B
MI
,
t
h
at
m
ig
h
t
b
e
m
is
s
ed
b
y
m
o
d
els tr
ea
tin
g
ea
ch
f
ea
tu
r
e
in
is
o
latio
n
.
Ho
wev
er
,
th
is
s
tr
en
g
th
is
also
th
e
s
o
u
r
ce
o
f
a
cr
itical
wea
k
n
ess
.
T
h
e
v
er
y
ar
ch
itectu
r
e
o
f
s
tan
d
ar
d
C
NNs
an
d
ANNs
is
f
u
n
d
am
en
tally
d
esig
n
ed
f
o
r
s
tatic,
p
o
in
t
-
in
-
tim
e
an
aly
s
is
.
T
h
ey
p
r
o
ce
s
s
a
s
in
g
le
“
s
n
ap
s
h
o
t
”
o
f
a
p
atien
t'
s
s
tate,
th
er
eb
y
co
m
n
eg
lectin
g
th
e
r
ich
,
lo
n
g
itu
d
in
al
n
ar
r
ativ
e
co
n
tain
ed
with
in
elec
tr
o
n
ic
h
ea
lth
r
ec
o
r
d
s
(
E
HR
s
)
.
Diab
etes
is
a
c
h
r
o
n
ic
,
p
r
o
g
r
ess
iv
e
d
is
ea
s
e;
a
p
atien
t'
s
ev
o
lv
in
g
lab
r
esu
lts
,
m
ed
icatio
n
ad
ju
s
tm
en
ts
,
an
d
f
lu
ctu
atin
g
g
lu
co
s
e
lev
els
o
v
er
tim
e
ar
e
p
ar
am
o
u
n
t
to
u
n
d
er
s
tan
d
in
g
th
eir
tr
ajec
to
r
y
an
d
p
r
ed
ictin
g
f
u
tu
r
e
o
u
tco
m
es.
A
m
o
d
el
th
at
o
n
ly
s
ee
s
th
e
m
o
s
t
r
ec
en
t
v
alu
es,
with
o
u
t
th
e
co
n
te
x
t
o
f
th
eir
h
is
to
r
ical
p
r
o
g
r
ess
io
n
,
is
o
p
er
atin
g
with
a
s
ev
er
e
in
f
o
r
m
atio
n
al
h
a
n
d
i
ca
p
.
T
h
is
lim
itatio
n
is
co
m
p
o
u
n
d
ed
wh
en
t
h
ese
m
o
d
els
f
ac
e
h
ig
h
-
d
im
en
s
io
n
al
f
ea
tu
r
e
s
p
ac
es,
wh
er
e
th
e
s
h
ee
r
n
u
m
b
er
o
f
v
ar
iab
les
(
e.
g
.
,
o
v
e
r
5
0
f
ea
tu
r
es
in
th
e
Diab
etes
1
3
0
-
US
d
ataset)
ca
n
lead
to
th
e
in
cr
ea
s
ed
d
im
e
n
s
io
n
ality
,
in
cr
ea
s
ed
co
m
p
u
tatio
n
al
co
s
t,
an
d
a
h
eig
h
ten
ed
r
is
k
o
f
o
v
er
f
itti
n
g
o
n
s
p
u
r
io
u
s
co
r
r
elatio
n
s
if
n
o
t
p
r
o
p
er
l
y
r
eg
u
lar
ized
o
r
r
e
d
u
ce
d
.
T
o
ad
d
r
ess
th
e
tem
p
o
r
al
asp
ec
t,
lo
n
g
s
h
o
r
t
-
te
r
m
m
e
m
o
r
y
(
L
STM
)
n
etwo
r
k
s
o
f
f
er
a
c
o
m
p
ellin
g
s
o
lu
tio
n
.
Yet,
th
is
in
t
r
o
d
u
ce
s
a
n
ew
s
et
o
f
c
h
allen
g
es.
T
h
e
ef
f
ec
tiv
e
ap
p
licatio
n
o
f
L
STM
s
o
f
ten
s
h
if
ts
th
e
b
u
r
d
e
n
o
f
co
m
p
lex
ity
f
r
o
m
th
e
m
o
d
el
its
elf
to
th
e
p
r
ep
r
o
ce
s
s
in
g
p
ip
elin
e.
I
t
r
eq
u
ir
es
m
e
ticu
lo
u
s
an
d
o
f
ten
co
m
p
le
x
f
ea
t
u
r
e
e
n
g
in
ee
r
in
g
to
s
tr
u
ctu
r
e
th
e
r
aw,
h
eter
o
g
en
eo
u
s
E
HR
d
ata
in
t
o
m
ea
n
in
g
f
u
l
tem
p
o
r
al
s
eq
u
en
ce
s
.
Fu
r
th
e
r
m
o
r
e,
wh
i
le
p
o
wer
f
u
l
f
o
r
te
m
p
o
r
al
p
a
tter
n
s
,
s
tan
d
alo
n
e
L
STM
s
ar
e
n
o
t
d
esig
n
ed
to
ef
f
icien
tl
y
ex
tr
ac
t
th
e
co
m
p
le
x
,
n
o
n
-
lin
ea
r
in
ter
ac
ti
o
n
s
b
et
wee
n
f
ea
tu
r
es
at
ea
ch
tim
e
s
tep
;
th
ey
ass
u
m
e
th
e
in
p
u
t f
ea
tu
r
es a
t e
ac
h
tim
estep
ar
e
alr
ea
d
y
o
p
tim
ally
in
f
o
r
m
a
tiv
e.
T
h
is
is
wh
er
e
t
h
e
s
y
n
er
g
is
tic
p
o
ten
tial
o
f
a
h
y
b
r
i
d
C
NN
-
L
STM
ar
ch
itectu
r
e
b
ec
o
m
es
ev
id
en
t,
a
n
ap
p
r
o
ac
h
th
at
h
as
y
ield
e
d
g
r
o
u
n
d
b
r
ea
k
in
g
r
esu
lts
in
ad
jace
n
t
m
ed
ical
d
o
m
ain
s
.
Fo
r
i
n
s
tan
ce
,
s
u
ch
h
y
b
r
id
s
h
av
e
b
ee
n
s
u
cc
ess
f
u
lly
d
ep
lo
y
ed
f
o
r
E
E
G
s
ig
n
al
class
if
ica
tio
n
,
wh
er
e
C
NNs
ex
tr
ac
t
s
p
atial
p
att
er
n
s
f
r
o
m
elec
tr
o
d
e
ar
r
ay
s
an
d
L
STM
s
m
o
d
el
th
e
t
em
p
o
r
al
ev
o
lu
tio
n
o
f
b
r
ain
ac
t
iv
ity
.
A
h
y
b
r
id
m
o
d
el
ca
n
,
in
t
h
eo
r
y
,
lev
e
r
ag
e
th
e
C
NN
co
m
p
o
n
en
t
to
p
er
f
o
r
m
a
u
to
m
atic
s
p
atial
f
ea
tu
r
e
lea
r
n
i
n
g
f
r
o
m
th
e
clin
ical
v
ar
ia
b
les
at
ea
ch
en
co
u
n
ter
,
ef
f
ec
tiv
ely
cr
ea
tin
g
a
r
ich
,
e
n
co
d
ed
r
ep
r
esen
tatio
n
o
f
th
e
p
atien
t's
s
tate
at
ea
ch
p
o
in
t
i
n
tim
e.
T
h
is
en
co
d
ed
s
eq
u
en
ce
is
th
en
f
ed
in
to
th
e
L
STM
,
wh
ich
lear
n
s
th
e
te
m
p
o
r
al
d
y
n
am
ics
an
d
p
r
o
g
r
e
s
s
io
n
b
etwe
en
th
ese
en
co
d
ed
s
tates.
T
h
is
en
d
-
to
-
e
n
d
lear
n
in
g
ap
p
r
o
ac
h
r
ep
r
esen
ts
a
s
ig
n
if
ican
t
ad
v
a
n
ce
m
en
t
o
v
er
m
o
d
els
th
at
r
eq
u
ir
e
s
ep
ar
ate,
m
a
n
u
al
f
ea
t
u
r
e
en
g
in
ee
r
in
g
f
o
r
th
e
tem
p
o
r
al
co
m
p
o
n
en
t
[
1
2
]
.
Similar
ly
,
in
g
en
o
m
ics,
th
ey
co
m
b
in
e
to
id
e
n
tify
s
p
atial
m
o
tifs
in
s
eq
u
en
ce
s
a
n
d
th
eir
tem
p
o
r
al
r
eg
u
latio
n
.
T
h
is
estab
lis
h
ed
s
u
cc
ess
s
tr
o
n
g
l
y
s
u
g
g
ests
its
u
n
tap
p
ed
p
o
te
n
tial f
o
r
d
ia
b
etes p
r
ed
ictio
n
[
1
3
]
.
T
h
er
ef
o
r
e,
wh
ile
th
e
in
d
iv
id
u
a
l
co
m
p
o
n
e
n
ts
(
C
NN
f
o
r
s
p
atiality
,
L
STM
f
o
r
tem
p
o
r
ality
)
a
r
e
k
n
o
wn
,
th
eir
in
teg
r
ated
a
p
p
licatio
n
to
d
iab
etes
p
r
ed
ictio
n
f
r
o
m
E
HR
d
ata
r
em
ain
s
a
r
elativ
ely
n
ascen
t
an
d
h
ig
h
-
p
o
ten
tial
r
esear
ch
av
en
u
e.
T
h
e
cr
itical
r
esear
ch
q
u
esti
o
n
ev
o
lv
es
f
r
o
m
s
im
p
ly
u
s
in
g
a
h
y
b
r
id
m
o
d
el
to
h
o
w
to
o
p
tim
ally
ar
ch
itect
an
d
f
ee
d
t
h
is
m
o
d
el.
Sp
ec
if
ically
,
h
o
w
to
r
ed
u
ce
th
e
h
ig
h
-
d
im
e
n
s
io
n
a
l
in
p
u
t
s
p
ac
e
to
its
m
o
s
t
in
f
o
r
m
ativ
e,
n
o
n
-
r
ed
u
n
d
an
t
elem
en
ts
to
en
h
an
ce
th
e
m
o
d
el'
s
ef
f
icien
cy
,
in
te
r
p
r
etab
ili
ty
,
an
d
p
er
f
o
r
m
an
ce
.
T
h
is
p
r
o
v
id
es
th
e
d
ir
ec
t
m
o
tiv
atio
n
f
o
r
in
teg
r
atin
g
an
ad
v
an
ce
d
f
ea
tu
r
e
s
elec
tio
n
m
ec
h
an
i
s
m
lik
e
DM
O)
as
a
p
r
ec
u
r
s
o
r
to
th
e
C
NN
-
L
STM
n
etwo
r
k
,
cr
ea
tin
g
a
p
o
wer
f
u
l
,
en
d
-
to
-
en
d
f
r
am
ewo
r
k
th
at
in
tellig
en
tly
s
elec
ts
f
ea
tu
r
es a
n
d
th
e
n
lear
n
s
b
o
th
t
h
eir
s
p
atial
an
d
tem
p
o
r
al
d
y
n
a
m
ics f
o
r
s
u
p
er
io
r
p
r
e
d
ictio
n
.
2
.
3
.
H
y
brid m
o
dels
(
wit
h f
e
a
t
ures select
io
n)
Featu
r
e
s
elec
tio
n
r
em
ain
s
a
cr
itical
ch
allen
g
e
in
d
iab
etes
p
r
ed
ictio
n
.
W
h
ile
m
eth
o
d
s
lik
e
ar
ith
m
etic
o
p
tim
izatio
n
alg
o
r
ith
m
an
d
m
em
etic
alg
o
r
ith
m
h
av
e
b
ee
n
ap
p
lied
[
1
4
]
,
th
ey
o
f
ten
s
u
f
f
er
f
r
o
m
p
r
em
atu
r
e
co
n
v
er
g
en
ce
o
r
h
ig
h
c
o
m
p
u
tat
io
n
al
co
m
p
lex
ity
.
T
h
e
DM
O
alg
o
r
ith
m
,
r
ec
e
n
tly
p
r
o
p
o
s
ed
b
y
[
1
5
]
,
o
f
f
er
s
s
ev
er
al
ad
v
an
tag
es:
a.
So
cial
h
ier
ar
ch
y
m
o
d
elin
g
:
m
im
ics
th
e
alp
h
a
-
le
d
g
r
o
u
p
s
tr
u
ctu
r
e
o
f
m
o
n
g
o
o
s
e
co
lo
n
ies
f
o
r
ef
f
icien
t
ex
p
lo
r
atio
n
.
b.
Dy
n
am
ic
b
alan
cin
g
: a
u
to
m
atic
ally
ad
ju
s
ts
ex
p
lo
r
atio
n
-
ex
p
lo
i
tatio
n
tr
ad
eo
f
f
d
u
r
in
g
o
p
tim
iz
atio
n
.
c.
C
o
m
p
u
tatio
n
al
ef
f
icien
c
y
: r
eq
u
ir
es f
ewe
r
iter
atio
n
s
th
an
c
o
m
p
ar
ab
le
alg
o
r
ith
m
s
d.
C
o
m
p
ar
ativ
e
s
tu
d
ies
h
av
e
s
h
o
wn
DM
O
o
u
tp
er
f
o
r
m
in
g
p
ar
ti
cle
s
war
m
o
p
tim
izatio
n
an
d
g
e
n
etic
alg
o
r
ith
m
s
o
n
b
en
ch
m
ar
k
p
r
o
b
lem
s
[
1
6
]
,
b
u
t its
ap
p
licatio
n
to
m
ed
ical
f
ea
tu
r
e
s
elec
tio
n
r
em
ain
s
lar
g
el
y
u
n
ex
p
lo
r
e
d
.
Mo
r
e
r
ec
en
t
a
p
p
licatio
n
s
co
n
ti
n
u
e
to
h
ig
h
lig
h
t
b
o
th
t
h
e
p
o
te
n
tial
an
d
th
e
p
itfa
lls
o
f
t
h
ese
m
eth
o
d
s
.
Fo
r
in
s
tan
ce
,
[
1
7
]
,
[
1
8
]
e
m
p
lo
y
e
d
h
y
p
e
r
p
ar
am
eter
-
tu
n
ed
en
s
em
b
le
m
eth
o
d
s
,
ac
h
iev
in
g
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
b
u
t
n
o
tin
g
s
ig
n
if
ican
t
s
en
s
itiv
ity
to
d
ata
q
u
ality
an
d
f
ea
t
u
r
e
s
elec
tio
n
.
T
h
e
s
tu
d
y
o
f
[
1
9
]
u
tili
ze
d
L
STM
s
to
m
o
d
el
p
atien
t
h
is
to
r
ies
f
o
r
p
r
ed
ictin
g
d
iab
etes
co
m
p
licatio
n
s
,
s
h
o
wca
s
in
g
th
eir
s
tr
en
g
th
in
ca
p
tu
r
in
g
lo
n
g
itu
d
in
al
p
atter
n
s
.
So
m
e
r
elate
d
s
tat
-
of
-
t
h
e
-
ar
t
wo
r
k
s
an
d
im
p
lem
en
tati
o
n
s
o
f
ML
an
d
DL
m
o
d
els
f
o
r
d
iab
etes
p
r
ed
ictio
n
ar
e
[
2
0
]
,
[
2
1
]
.
T
h
e
s
tu
d
y
o
f
[
2
2
]
p
r
o
v
id
e
d
a
co
m
p
r
eh
e
n
s
iv
e
s
u
r
v
ey
c
o
n
clu
d
in
g
t
h
at
wh
ile
class
ical
ML
i
s
ef
f
ec
tiv
e,
its
ce
ilin
g
is
l
im
ited
with
o
u
t
ad
v
an
ce
d
f
ea
t
u
r
e
en
g
i
n
ee
r
i
n
g
o
r
in
teg
r
atio
n
with
m
o
r
e
p
o
wer
f
u
l
lear
n
i
n
g
p
ar
ad
ig
m
s
.
Fo
r
ex
am
p
le,
th
e
s
t
u
d
y
o
f
[
2
3
]
c
o
m
b
in
e
d
f
ea
t
u
r
e
s
elec
tio
n
with
an
en
s
em
b
le
o
f
cl
ass
if
ier
s
,
wh
ile
th
e
s
tu
d
y
o
f
[
2
4
]
ex
p
lo
r
ed
th
e
s
y
n
er
g
y
b
etwe
en
o
p
tim
izatio
n
alg
o
r
ith
m
s
an
d
n
eu
r
al
n
etwo
r
k
s
.
Similar
ly
,
[
2
5
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
6
,
Decem
b
e
r
20
25
:
5
5
5
5
-
5
5
6
9
5558
d
em
o
n
s
tr
ated
th
at
wh
ile
m
o
d
els
lik
e
R
F
ca
n
ac
h
iev
e
h
ig
h
ac
cu
r
ac
y
(
~9
4
%),
th
eir
p
er
f
o
r
m
an
ce
is
h
ea
v
ily
d
ep
en
d
e
n
t
o
n
th
e
d
ataset'
s
ch
ar
ac
ter
is
tics
an
d
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
.
A
cr
it
ical
lim
ita
tio
n
r
em
ain
s
th
eir
in
h
er
en
t
in
ab
ilit
y
to
a
u
to
n
o
m
o
u
s
ly
lear
n
co
m
p
lex
,
h
ier
ar
c
h
ical
f
ea
tu
r
e
in
ter
ac
tio
n
s
f
r
o
m
r
aw
d
ata,
r
ely
in
g
i
n
s
tead
o
n
ex
p
er
t
-
d
r
iv
en
f
ea
tu
r
e
c
u
r
atio
n
.
R
ec
o
g
n
izin
g
th
e
s
tr
en
g
th
s
o
f
d
if
f
er
e
n
t
p
ar
a
d
ig
m
s
,
r
ec
e
n
t
r
esear
ch
h
as
s
h
if
ted
to
war
d
s
h
y
b
r
id
m
o
d
els
th
at
in
t
eg
r
ate
f
ea
tu
r
e
s
elec
tio
n
,
o
p
tim
izatio
n
alg
o
r
ith
m
s
,
an
d
d
ee
p
le
ar
n
in
g
.
I
n
[
2
6
]
,
t
h
e
s
tu
d
y
p
r
o
v
id
e
d
a
co
m
p
r
eh
e
n
s
iv
e
r
ev
iew,
co
n
cl
u
d
in
g
th
at
h
y
b
r
id
m
o
d
els
c
o
n
s
is
ten
tly
o
u
t
p
er
f
o
r
m
s
tan
d
alo
n
e
class
if
ier
s
.
R
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
(
R
NNs),
p
ar
ticu
lar
ly
lo
n
g
s
h
o
r
t
-
t
er
m
m
em
o
r
y
(
L
STM
)
an
d
g
ate
d
r
ec
u
r
r
e
n
t
u
n
it
(
GR
U)
n
etwo
r
k
s
,
ar
e
n
at
u
r
ally
s
u
ited
f
o
r
tem
p
o
r
al
d
at
a,
s
u
ch
as
p
atien
t
E
HR
s
eq
u
e
n
ce
s
.
T
h
e
s
tu
d
ies
o
f
[
2
7
]
,
[
2
8
]
f
u
r
th
e
r
d
em
o
n
s
tr
ated
th
at
an
L
STM
m
o
d
el
with
a
tten
tio
n
m
ec
h
an
is
m
s
co
u
ld
id
en
tify
cr
itical
tim
e
p
o
in
ts
in
a
p
atien
t'
s
h
is
to
r
y
f
o
r
p
r
ed
ictio
n
.
Fu
r
th
er
m
o
r
e,
[
2
9
]
p
o
in
ted
o
u
t
th
at
d
ee
p
lea
r
n
in
g
m
o
d
els
ar
e
h
ig
h
l
y
s
u
s
ce
p
tib
le
to
p
e
r
f
o
r
m
an
ce
d
e
g
r
ad
atio
n
ca
u
s
ed
b
y
class
im
b
alan
ce
p
r
e
v
alen
t
in
m
e
d
ical
d
a
tasets
lik
e
Diab
etes
130
-
US,
o
f
ten
r
e
q
u
ir
in
g
s
o
p
h
i
s
ticated
s
am
p
lin
g
tech
n
iq
u
es.
2
.
4
.
Resea
rc
h
g
a
ps
Ou
r
co
m
p
r
eh
en
s
iv
e
r
e
v
iew
o
f
t
h
e
liter
atu
r
e
id
en
tifie
s
th
r
ee
p
e
r
s
is
ten
t
an
d
in
ter
co
n
n
ec
ted
r
esear
ch
g
ap
s
th
at
h
av
e
lim
ited
th
e
p
e
r
f
o
r
m
an
ce
an
d
g
e
n
er
aliza
b
ilit
y
o
f
p
r
ev
i
o
u
s
d
iab
etes
p
r
ed
ictio
n
m
o
d
els:
i
)
lim
ited
tem
p
o
r
al
m
o
d
elin
g
;
ii
)
s
u
b
o
p
ti
m
al
f
ea
tu
r
e
s
elec
tio
n
; a
n
d
iii
)
ar
ch
itectu
r
al
co
n
s
tr
ain
ts
.
T
h
ese
g
ap
s
in
d
icate
th
at
ex
is
tin
g
m
o
d
els
o
f
ten
s
tr
u
g
g
le
to
ca
p
tu
r
e
th
e
d
y
n
am
ic
n
atu
r
e
o
f
p
atien
t
h
ea
lth
r
ec
o
r
d
s
,
in
ad
e
q
u
ately
em
p
h
asize
th
e
id
en
tific
atio
n
o
f
th
e
m
o
s
t
in
f
o
r
m
ativ
e
f
ea
tu
r
es,
an
d
d
e
p
en
d
o
n
r
i
g
id
ar
ch
itectu
r
al
d
esig
n
s
th
at
r
ed
u
ce
ad
ap
tab
ilit
y
.
E
ac
h
o
f
th
ese
g
a
p
s
is
d
is
cu
s
s
ed
in
d
etail
to
s
h
o
w
h
o
w
th
ey
co
n
s
tr
ain
p
r
e
d
ictiv
e
p
er
f
o
r
m
an
ce
an
d
to
o
u
tlin
e
d
ir
ec
ti
o
n
s
f
o
r
m
o
r
e
ef
f
ec
tiv
e
m
o
d
el
d
ev
elo
p
m
en
t.
First,
a
p
r
ed
o
m
i
n
an
t
g
a
p
is
th
e
wid
esp
r
ea
d
n
e
g
lect
o
f
tem
p
o
r
al
d
y
n
am
ics.
T
h
e
m
ajo
r
ity
o
f
ex
is
tin
g
ap
p
r
o
ac
h
es,
in
clu
d
i
n
g
m
o
s
t
tr
ad
itio
n
al
m
ac
h
i
n
e
lear
n
i
n
g
m
o
d
els
(
e.
g
.
,
SVM,
r
a
n
d
o
m
f
o
r
est
)
an
d
e
v
en
m
a
n
y
s
tan
d
ar
d
d
ee
p
lea
r
n
in
g
m
o
d
e
ls
(
e.
g
.
,
ML
P,
b
asic
C
NN)
,
t
r
ea
t
co
m
p
lex
p
atien
t
h
is
to
r
ie
s
as
s
tatic,
is
o
lated
s
n
ap
s
h
o
ts
[
1
9
]
.
T
h
is
is
a
cr
itic
al
o
v
er
s
ig
h
t
f
o
r
p
r
o
g
r
ess
iv
e
co
n
d
itio
n
lik
e
d
iab
etes
m
ellitu
s
,
wh
er
e
th
e
tr
ajec
to
r
y
o
f
b
i
o
m
ar
k
e
r
s
s
u
ch
as
Hb
A
1
c,
f
asti
n
g
g
lu
c
o
s
e,
an
d
m
e
d
icatio
n
ch
a
n
g
es
o
v
er
tim
e
co
n
tain
s
in
v
al
u
ab
le
p
r
o
g
n
o
s
tic
in
f
o
r
m
atio
n
.
B
y
f
a
ilin
g
to
m
o
d
el
th
ese
lo
n
g
itu
d
i
n
al
s
eq
u
en
ce
s
,
th
ese
ap
p
r
o
ac
h
es
d
is
ca
r
d
a
cr
u
cial
d
im
en
s
io
n
o
f
th
e
clin
ical
n
a
r
r
a
tiv
e,
in
ev
itab
ly
ca
p
p
in
g
th
eir
p
r
ed
ictiv
e
p
o
ten
tial a
n
d
clin
ical
u
tili
ty
.
Seco
n
d
,
th
e
p
r
o
ce
s
s
o
f
f
ea
tu
r
e
s
elec
tio
n
r
em
ain
s
a
s
ig
n
if
ican
t
b
o
ttlen
ec
k
.
W
h
ile
tech
n
iq
u
es
lik
e
p
r
in
cip
al
co
m
p
o
n
e
n
t
an
aly
s
is
(
PC
A)
,
ch
i
-
s
q
u
ar
e
test
s
,
an
d
ev
en
m
etah
eu
r
is
tics
lik
e
g
en
etic
alg
o
r
ith
m
s
(
GA)
o
r
p
ar
ticle
s
war
m
o
p
tim
izatio
n
(
PS
O)
ar
e
co
m
m
o
n
l
y
em
p
l
o
y
e
d
,
th
ey
ar
e
o
f
ten
s
u
b
o
p
tim
al.
T
h
ese
m
eth
o
d
s
ca
n
s
u
f
f
er
f
r
o
m
p
r
em
atu
r
e
c
o
n
v
e
r
g
en
ce
,
g
et
tr
ap
p
ed
in
lo
ca
l
o
p
tim
a,
o
r
lack
a
m
ec
h
an
is
m
to
ef
f
icien
tly
b
alan
ce
th
e
ex
p
lo
r
atio
n
o
f
n
ew
f
ea
tu
r
e
s
u
b
s
ets
with
th
e
ex
p
lo
itatio
n
o
f
k
n
o
wn
g
o
o
d
o
n
es.
C
o
n
s
eq
u
en
tly
,
t
h
ey
f
r
eq
u
en
tly
y
ield
f
ea
tu
r
e
s
u
b
s
ets
th
at
co
n
tain
r
ed
u
n
d
an
cies
o
r
ir
r
elev
a
n
t
v
ar
iab
les,
wh
ich
ca
n
in
tr
o
d
u
ce
n
o
is
e,
in
c
r
ea
s
e
co
m
p
u
tatio
n
al
o
v
er
h
ea
d
,
an
d
u
ltima
tely
d
eg
r
ad
e
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
d
o
wn
s
tr
ea
m
class
if
ier
.
T
h
er
e
is
a
clea
r
n
ee
d
f
o
r
a
m
o
r
e
r
o
b
u
s
t
an
d
in
tellig
en
t
f
ea
tu
r
e
s
elec
tio
n
s
tr
ateg
y
th
at
is
d
ir
ec
tly
o
p
tim
ized
f
o
r
th
e
s
p
ec
if
ic
p
r
ed
ictiv
e
task
.
T
h
ir
d
,
th
er
e
ar
e
f
u
n
d
am
e
n
tal
ar
ch
itectu
r
al
co
n
s
tr
ain
ts
in
co
m
m
o
n
ly
u
s
ed
class
if
ier
s
.
Sim
p
le
m
o
d
els
lik
e
lo
g
is
tic
r
eg
r
ess
io
n
o
r
d
ec
is
io
n
tr
ee
s
lack
th
e
ca
p
ac
ity
t
o
m
o
d
el
co
m
p
lex
n
o
n
-
lin
ea
r
r
elatio
n
s
h
ip
s
.
W
h
ile
m
o
r
e
p
o
wer
f
u
l,
s
tan
d
al
o
n
e
m
o
d
els
lik
e
C
NNs
o
r
L
STM
s
h
av
e
th
eir
o
wn
lim
itatio
n
s
:
C
NNs
ar
e
ad
ep
t
at
id
en
tify
in
g
l
o
ca
l
s
p
atial
p
atter
n
s
an
d
in
ter
ac
tio
n
s
b
etwe
en
f
e
atu
r
es
at
a
s
in
g
le
p
o
in
t
in
t
im
e
b
u
t
ar
e
a
g
n
o
s
tic
to
s
eq
u
en
ce
,
wh
er
ea
s
L
STM
s
ex
ce
l
at
m
o
d
elin
g
tem
p
o
r
al
s
eq
u
en
ce
s
b
u
t
ar
e
n
o
t
d
esig
n
ed
t
o
ef
f
ic
ien
tly
ex
tr
ac
t
co
m
p
lex
s
p
atial
f
ea
tu
r
e
h
ie
r
ar
ch
ies
f
r
o
m
a
s
tatic
in
p
u
t
v
ec
t
o
r
.
A
n
ar
c
h
itectu
r
e
th
at
ca
n
s
ea
m
less
ly
in
teg
r
ate
th
ese
two
ca
p
ab
ilit
ies
—
s
p
atia
l
f
ea
tu
r
e
lear
n
in
g
a
n
d
tem
p
o
r
al
s
eq
u
en
ce
m
o
d
elin
g
—
is
th
e
r
ef
o
r
e
n
ec
ess
ar
y
to
f
u
lly
lev
er
a
g
e
th
e
in
f
o
r
m
a
tio
n
co
n
tain
ed
with
in
m
u
ltid
im
e
n
s
io
n
al
E
HR
d
ata.
Ou
r
p
r
o
p
o
s
ed
DM
O
-
C
NN
-
L
ST
M
m
o
d
el
is
ar
ch
itected
s
p
ec
if
ica
lly
to
b
r
id
g
e
th
ese
cr
itical
g
ap
s
th
r
o
u
g
h
a
n
o
v
el
in
teg
r
ati
o
n
o
f
b
io
-
in
s
p
ir
e
d
o
p
tim
izatio
n
an
d
h
y
b
r
id
d
ee
p
l
ea
r
n
in
g
.
T
o
a
d
d
r
ess
Ga
p
1
(
t
em
p
o
r
al
m
o
d
eli
n
g
)
,
we
em
p
l
o
y
a
h
y
b
r
i
d
C
NN
-
L
STM
a
r
c
h
it
ec
t
u
r
e
.
T
h
e
C
NN
la
y
e
r
s
f
i
r
s
t
a
ct
as
a
u
t
o
m
ati
c
f
ea
tu
r
e
e
x
tr
a
ct
o
r
s
,
l
ea
r
n
i
n
g
n
o
n
-
li
n
ea
r
s
p
atia
l
c
o
r
r
ela
ti
o
n
s
a
n
d
h
i
er
a
r
c
h
i
es
wit
h
i
n
t
h
e
cli
n
i
ca
l
f
e
at
u
r
es
o
f
ea
c
h
in
d
i
v
id
u
a
l
p
ati
en
t
en
c
o
u
n
t
er
.
T
h
e
o
u
t
p
u
t
o
f
t
h
is
s
p
at
ial
an
al
y
s
is
is
t
h
e
n
f
e
d
a
s
a
s
e
q
u
e
n
t
ial
i
n
p
u
t
to
th
e
L
STM
la
y
er
,
w
h
i
c
h
is
s
p
ec
i
f
i
ca
l
ly
d
esi
g
n
e
d
t
o
l
ea
r
n
t
h
e
lo
n
g
-
te
r
m
d
e
p
e
n
d
e
n
cies
a
n
d
te
m
p
o
r
a
l
p
at
te
r
n
s
b
e
twe
en
t
h
es
e
e
n
c
o
d
e
d
e
n
c
o
u
n
t
e
r
s
,
e
f
f
ec
ti
v
el
y
m
o
d
e
lin
g
t
h
e
p
ati
en
t'
s
d
is
e
ase
p
r
o
g
r
ess
io
n
o
v
er
ti
m
e
.
T
o
ad
d
r
ess
Gap
2
(
s
u
b
o
p
tim
al
f
ea
tu
r
e
s
elec
tio
n
)
,
we
in
te
g
r
ate
th
e
d
war
f
m
o
n
g
o
o
s
e
o
p
tim
izatio
n
(
DM
O)
alg
o
r
ith
m
as
a
n
in
telli
g
en
t
p
r
e
-
p
r
o
ce
s
s
in
g
s
tep
.
Un
li
k
e
tr
ad
itio
n
al
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
,
DM
O's
s
o
cial
h
ier
ar
ch
y
an
d
d
y
n
am
ic
f
o
r
ag
in
g
b
eh
a
v
io
r
p
r
o
v
i
d
e
a
s
u
p
er
io
r
m
ec
h
a
n
is
m
f
o
r
n
av
i
g
atin
g
th
e
co
m
p
lex
s
ea
r
ch
s
p
ac
e
o
f
p
o
te
n
tial
f
ea
t
u
r
e
s
u
b
s
ets.
I
t
ef
f
icien
tly
b
al
an
ce
s
ex
p
lo
r
atio
n
an
d
ex
p
lo
it
atio
n
to
id
e
n
tify
a
p
ar
s
im
o
n
io
u
s
s
et
o
f
h
ig
h
ly
p
r
ed
ictiv
e
f
ea
tu
r
es,
d
ir
ec
tly
o
p
ti
m
izin
g
f
o
r
th
e
v
alid
atio
n
ac
c
u
r
ac
y
o
f
th
e
C
NN
-
L
STM
m
o
d
el
its
elf
,
th
u
s
en
s
u
r
in
g
th
e
s
elec
ted
f
ea
tu
r
es a
r
e
m
ax
im
ally
r
elev
an
t
f
o
r
t
h
e
f
in
al
p
r
ed
ictio
n
task
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
DMO
-
CNN
-
LS
T
M fr
a
mewo
r
k
fo
r
fea
tu
r
e
s
elec
tio
n
a
n
d
d
i
a
b
etes
…
(
Mu
ta
s
em
K
.
A
ls
ma
d
i
)
5559
T
o
ad
d
r
ess
Gap
3
(
ar
ch
itect
u
r
al
co
n
s
tr
ain
ts
)
,
th
e
en
tire
f
r
am
ewo
r
k
is
d
esig
n
ed
as
an
en
d
-
to
-
e
n
d
p
ip
elin
e
th
at
s
y
n
er
g
izes th
e
s
tr
en
g
th
s
o
f
its
co
m
p
o
n
e
n
ts
.
T
h
e
DM
O
alg
o
r
ith
m
h
an
d
les th
e
h
ig
h
-
d
im
e
n
s
io
n
ality
an
d
n
o
is
e,
th
e
C
NN
h
an
d
les
s
p
atial
f
ea
tu
r
e
lear
n
in
g
,
an
d
th
e
L
STM
h
an
d
les
tem
p
o
r
al
m
o
d
e
lin
g
.
T
h
is
co
h
esiv
e
s
tr
u
ctu
r
e
m
o
v
es
b
ey
o
n
d
s
im
p
l
e
m
o
d
el
s
tack
in
g
t
o
cr
ea
te
a
u
n
if
ied
s
y
s
tem
ca
p
a
b
le
o
f
s
im
u
ltan
eo
u
s
ly
lear
n
i
n
g
f
r
o
m
b
o
th
th
e
s
p
atial
an
d
te
m
p
o
r
al
d
im
en
s
io
n
s
o
f
th
e
d
ata,
th
er
e
b
y
o
v
er
c
o
m
in
g
th
e
in
h
er
en
t
lim
itatio
n
s
o
f
s
im
p
ler
o
r
s
tan
d
alo
n
e
class
if
ier
s
.
B
y
co
n
f
r
o
n
tin
g
t
h
ese
th
r
ee
g
a
p
s
d
ir
ec
tly
,
o
u
r
p
r
o
p
o
s
ed
m
o
d
el
o
f
f
er
s
a
m
o
r
e
s
o
p
h
is
ticated
,
r
o
b
u
s
t,
an
d
clin
ically
r
elev
an
t
f
r
am
ewo
r
k
f
o
r
in
tellig
en
t
d
iab
etes
p
r
ed
ictio
n
.
Un
lik
e
p
r
ev
io
u
s
ap
p
r
o
ac
h
es,
th
e
m
o
d
el
in
teg
r
ates
tem
p
o
r
al
d
y
n
am
ics,
o
p
tim
ized
f
ea
tu
r
e
s
elec
tio
n
,
an
d
f
lex
ib
le
ar
c
h
itectu
r
al
d
esig
n
s
to
en
s
u
r
e
b
o
th
ac
cu
r
ac
y
an
d
g
e
n
er
aliza
b
ilit
y
.
T
h
is
co
m
p
r
eh
e
n
s
iv
e
ap
p
r
o
ac
h
en
h
an
ce
s
p
r
e
d
ictiv
e
p
er
f
o
r
m
a
n
ce
an
d
s
tr
en
g
th
e
n
s
th
e
m
o
d
el’
s
p
o
ten
tial to
p
r
o
v
i
d
e
m
ea
n
in
g
f
u
l su
p
p
o
r
t in
r
ea
l
-
wo
r
ld
clin
ical
s
ettin
g
s
.
3.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
e
m
eth
o
d
o
lo
g
y
o
f
th
is
s
tu
d
y
in
teg
r
ates
an
in
tellig
en
t
f
ea
tu
r
e
s
elec
tio
n
alg
o
r
ith
m
DM
O
with
a
h
y
b
r
id
d
ee
p
lear
n
in
g
ar
c
h
itectu
r
e,
C
NN
-
L
STM
,
to
en
h
an
ce
d
iab
etes
p
r
ed
ictio
n
ac
cu
r
ac
y
.
T
h
e
p
r
o
ce
s
s
is
d
iv
id
ed
in
t
o
f
iv
e
m
ain
s
tag
es: d
ata
p
r
ep
r
o
c
ess
in
g
,
f
ea
tu
r
e
s
elec
tio
n
,
m
o
d
el
ar
ch
itectu
r
e
d
esig
n
,
tr
ain
in
g
an
d
v
alid
atio
n
,
a
n
d
co
m
p
ar
ativ
e
e
v
alu
atio
n
.
See
F
ig
u
r
e
1
a
n
d
Alg
o
r
ith
m
1.
Fig
u
r
e
1
.
Me
th
o
d
o
lo
g
y
o
f
th
e
p
r
o
p
o
s
ed
DM
O
-
C
NN
-
L
STM
m
o
d
el
f
o
r
d
iab
etes p
r
ed
ictio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
6
,
Decem
b
e
r
20
25
:
5
5
5
5
-
5
5
6
9
5560
Alg
o
r
ith
m
1
.
DM
O
f
o
r
f
ea
tu
r
e
s
elec
tio
n
Input:
-
D: Dataset with N features
-
MaxIter: Maximum number of iterations
-
PopSize: Number of mongooses (solutions)
-
Fitness(): Fitness function (CNN
-
LSTM validation accuracy)
Output:
-
BestFeatureSubset
Begin
1.
Initialize
population of mongooses (random binary vectors of N features)
2.
Evaluate
fitness of each mongoose using CNN
-
LSTM accuracy
3.
Store
the best solution as AlphaMongoose
For
iter
=
1 to MaxIter do
For
each mongoose i in population do
-
Perform random movement (exploration)
-
If better fitness, update AlphaMongoose
End
for
For
each mongoose i do
-
Local search near AlphaMongoose (exploitation)
-
Update if fitness improves
End
for
End
for
Return
AlphaMongoose as BestFeatureSubset
End
T
h
e
p
r
o
p
o
s
ed
DM
O
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
with
th
e
C
N
N
-
L
STM
clas
s
if
ier
o
f
f
er
s
s
ev
er
al
ad
v
an
tag
es.
First,
DM
O
en
ab
les
f
ea
tu
r
e
s
e
lectio
n
at
an
ea
r
ly
s
tag
e
b
y
elim
in
atin
g
r
ed
u
n
d
a
n
t
f
ea
tu
r
es,
th
u
s
r
ed
u
cin
g
n
o
is
e,
co
m
p
u
tatio
n
al
co
m
p
lex
ity
,
an
d
th
e
r
is
k
o
f
o
v
er
f
i
ttin
g
.
Sec
o
n
d
,
th
e
h
y
b
r
id
ar
ch
itectu
r
e
o
f
f
er
s
a
b
alan
ce
o
f
lear
n
in
g
,
wh
er
e
t
h
e
C
NN
co
m
p
o
n
e
n
t
ef
f
icien
tly
ex
tr
ac
ts
lo
ca
l
f
ea
tu
r
e
p
atter
n
s
an
d
th
e
L
STM
co
m
p
o
n
en
t
ca
p
tu
r
es
tem
p
o
r
al
d
ep
en
d
en
ci
es,
allo
win
g
th
e
m
o
d
el
to
lear
n
b
o
th
s
tatic
an
d
d
y
n
am
ic
c
h
ar
a
cter
is
t
ics
o
f
m
ed
ical
f
ea
tu
r
es.
T
h
ir
d
,
th
e
u
s
e
o
f
a
co
m
p
ac
t
an
d
d
is
cr
im
in
ativ
e
f
ea
tu
r
e
s
u
b
s
et
en
h
an
c
es
g
en
er
aliza
tio
n
,
im
p
r
o
v
in
g
r
o
b
u
s
tn
ess
ac
r
o
s
s
h
eter
o
g
en
e
o
u
s
m
ed
ical
d
atasets
.
T
h
e
li
g
h
tweig
h
t
C
NN
-
L
STM
ev
alu
atio
n
with
DM
O
co
n
f
ir
m
s
tr
ac
tab
ilit
y
ev
e
n
in
h
ig
h
-
d
im
e
n
s
io
n
al
s
ea
r
ch
s
p
ac
es,
wh
ile
p
o
o
lin
g
o
p
e
r
atio
n
s
r
ed
u
ce
co
m
p
u
tatio
n
al
lo
ad
.
Ad
d
itio
n
ally
,
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
is
s
ig
n
if
ican
tly
e
n
h
an
ce
d
,
as
th
e
in
ter
ac
tio
n
b
et
wee
n
DM
O
-
d
r
iv
en
f
ea
tu
r
e
s
elec
tio
n
an
d
h
y
b
r
id
C
NN
-
L
STM
class
if
icat
io
n
im
p
r
o
v
es o
v
er
all
p
er
f
o
r
m
an
ce
.
3
.
1
.
Da
t
a
s
et
a
nd
prepro
ce
s
s
i
ng
T
h
e
d
ataset
u
s
ed
f
o
r
th
is
s
t
u
d
y
is
th
e
well
-
estab
lis
h
ed
Diab
etes
1
3
0
-
US
h
o
s
p
itals
[
3
0
]
d
ataset,
co
m
p
r
is
in
g
o
v
er
1
0
0
,
0
0
0
r
ec
o
r
d
s
co
llected
o
v
er
a
1
0
-
y
ea
r
p
er
io
d
a
n
d
5
5
attr
ib
u
tes,
in
clu
d
in
g
d
em
o
g
r
a
p
h
ics,
d
iag
n
o
s
es,
lab
r
esu
lts
,
an
d
h
o
s
p
ital
o
u
tco
m
es.
Af
ter
r
em
o
v
in
g
id
en
tifie
r
s
s
u
ch
as
en
co
u
n
ter
_
id
an
d
p
atien
t_
n
b
r
,
we
p
er
f
o
r
m
e
d
p
r
ep
r
o
ce
s
s
in
g
to
clea
n
an
d
s
tan
d
ar
d
ize
th
e
d
at
a.
All
m
is
s
in
g
v
alu
es
an
d
in
co
n
s
is
ten
t
en
tr
ies
wer
e
r
ep
lace
d
u
s
in
g
ap
p
r
o
p
r
iate
im
p
u
tatio
n
s
tr
ateg
ies
o
r
th
e
af
f
ec
t
ed
co
lu
m
n
s
wer
e
d
r
o
p
p
ed
if
m
o
r
e
th
a
n
5
0
%
o
f
th
e
d
ata
was
m
is
s
in
g
.
C
ateg
o
r
ical
attr
ib
u
tes
wer
e
en
co
d
ed
u
s
in
g
L
a
b
el
E
n
co
d
i
n
g
,
an
d
th
e
co
m
p
lete
d
ataset
wa
s
n
o
r
m
alize
d
u
s
in
g
Min
-
Ma
x
Sc
alin
g
to
en
s
u
r
e
f
ea
tu
r
e
r
a
n
g
es
wer
e
co
n
s
is
ten
t,
wh
ich
is
cr
u
c
ial
f
o
r
co
n
v
er
g
e
n
ce
in
n
eu
r
al
n
etwo
r
k
s
.
3
.
2
.
F
e
a
t
ure
s
elec
t
io
n us
ing
DM
O
Featu
r
e
s
elec
tio
n
is
a
cr
itical
p
h
ase
in
t
h
e
m
eth
o
d
o
lo
g
y
,
a
s
ir
r
elev
an
t
o
r
r
ed
u
n
d
a
n
t
attr
i
b
u
tes
ca
n
d
eg
r
ad
e
m
o
d
el
p
er
f
o
r
m
an
ce
a
n
d
in
cr
ea
s
e
c
o
m
p
u
tatio
n
al
co
s
t.
T
o
ad
d
r
ess
th
is
,
we
ap
p
lied
t
h
e
DM
O
alg
o
r
ith
m
,
a
m
etah
eu
r
is
tic
in
s
p
ir
ed
b
y
th
e
co
o
p
er
ativ
e
h
u
n
tin
g
an
d
co
m
m
u
n
icatio
n
s
tr
ateg
ies o
f
d
war
f
m
o
n
g
o
o
s
es.
DM
O
in
itializes a
p
o
p
u
latio
n
(
s
ize
=
20
-
1
0
0
)
o
f
r
a
n
d
o
m
f
ea
tu
r
e
s
u
b
s
ets,
wh
er
e
ea
ch
in
d
iv
id
u
al
is
en
co
d
ed
as a
b
in
ar
y
v
ec
to
r
(
1
=
s
elec
ted
,
0
=
ig
n
o
r
e
d
)
r
ep
r
esen
tin
g
th
e
in
cl
u
s
io
n
o
r
ex
clu
s
io
n
o
f
f
ea
tu
r
es.
T
h
e
f
i
tn
ess
o
f
ea
ch
s
u
b
s
et
is
ev
alu
ated
u
s
in
g
th
e
class
if
icatio
n
ac
cu
r
ac
y
o
f
a
lig
h
tweig
h
t
C
NN
-
L
STM
m
o
d
el
tr
ain
ed
o
v
er
th
r
ee
e
p
o
ch
s
.
DM
O
em
p
lo
y
s
a
s
to
ch
asti
c
eli
te
-
b
ased
s
ea
r
ch
s
tr
ateg
y
,
b
alan
cin
g
ex
p
lo
r
atio
n
an
d
ex
p
lo
itati
o
n
as
it
u
p
d
ates
th
e
p
o
p
u
latio
n
o
v
er
m
u
ltip
le
iter
at
io
n
s
(
e.
g
.
,
5
-
1
0
0
)
.
T
h
e
b
est
-
p
e
r
f
o
r
m
in
g
f
ea
tu
r
e
s
u
b
s
et
is
s
ele
cted
f
o
r
f
in
al
m
o
d
el
tr
ain
in
g
,
ty
p
ically
co
m
p
r
is
in
g
1
0
to
2
0
attr
ib
u
tes.
3
.
3
.
CNN
-
L
ST
M
a
rc
hite
ct
u
re
f
o
r
cla
s
s
if
ica
t
io
n
T
o
class
if
y
th
e
o
p
tim
ized
f
ea
tu
r
e
s
u
b
s
et,
we
d
esig
n
ed
a
h
y
b
r
id
C
NN
-
L
STM
m
o
d
el.
T
h
e
C
NN
lay
er
s
ar
e
r
esp
o
n
s
ib
le
f
o
r
e
x
tr
ac
tin
g
lo
ca
l
s
p
atial
p
atter
n
s
an
d
f
ea
tu
r
e
in
ter
ac
tio
n
s
,
wh
ile
th
e
L
ST
M
u
n
its
ar
e
d
esig
n
e
d
to
ca
p
tu
r
e
lo
n
g
-
ter
m
d
ep
e
n
d
en
cies
an
d
s
eq
u
en
tial
r
elatio
n
s
h
ip
s
,
wh
ich
ar
e
esp
ec
ially
u
s
ef
u
l
f
o
r
m
ed
ical
f
ea
tu
r
es.
T
h
e
ar
c
h
itectu
r
e
in
cl
u
d
es
o
n
e
1
D
c
o
n
v
o
lu
tio
n
al
lay
er
with
R
eL
U
ac
tiv
atio
n
,
f
o
llo
wed
b
y
m
a
x
p
o
o
lin
g
to
r
ed
u
ce
d
im
e
n
s
io
n
ality
.
T
h
e
o
u
tp
u
t
is
th
en
p
ass
ed
in
to
an
L
STM
lay
er
with
6
4
m
em
o
r
y
ce
lls
,
f
o
llo
wed
b
y
a
d
en
s
e
lay
er
with
a
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
f
o
r
b
in
ar
y
cl
ass
if
icatio
n
(
d
iab
etic
o
r
n
o
n
-
d
iab
etic)
.
T
h
e
m
o
d
el
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
DMO
-
CNN
-
LS
T
M fr
a
mewo
r
k
fo
r
fea
tu
r
e
s
elec
tio
n
a
n
d
d
i
a
b
etes
…
(
Mu
ta
s
em
K
.
A
ls
ma
d
i
)
5561
is
co
m
p
iled
u
s
in
g
t
h
e
Ad
am
o
p
tim
izer
with
a
b
in
a
r
y
c
r
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
an
d
tr
ain
e
d
f
o
r
2
0
–
5
0
e
p
o
ch
s
d
ep
en
d
i
n
g
o
n
th
e
ex
p
er
im
en
t.
Fig
u
r
e
2
illu
s
tr
ates th
e
o
v
er
all
ar
ch
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
C
NN
-
L
STM
m
o
d
el,
h
ig
h
lig
h
tin
g
th
e
s
eq
u
en
tial
f
lo
w
f
r
o
m
t
h
e
in
p
u
t
lay
er
th
r
o
u
g
h
co
n
v
o
lu
tio
n
,
p
o
o
lin
g
,
an
d
r
e
cu
r
r
en
t
la
y
er
s
,
an
d
f
in
ally
to
th
e
d
en
s
e
s
ig
m
o
id
-
a
ctiv
ated
o
u
tp
u
t f
o
r
b
in
ar
y
class
if
icatio
n
.
Fig
u
r
e
2
.
Hy
b
r
id
C
NN
-
L
STM
ar
ch
itectu
r
e
3
.
4
.
T
ra
ini
ng
a
nd
ev
a
lua
t
io
n str
a
t
eg
y
T
h
e
d
ataset
is
s
p
lit
in
to
7
0
%
tr
ain
in
g
an
d
3
0
%
test
in
g
p
ar
titi
o
n
s
.
Per
f
o
r
m
a
n
ce
is
ev
alu
ated
u
s
in
g
k
ey
m
etr
ics
s
u
ch
as
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
R
O
C
-
AUC,
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
,
a
n
d
m
ea
n
ab
s
o
lu
te
er
r
o
r
(
MA
E
)
.
I
n
ad
d
itio
n
,
c
o
n
f
u
s
io
n
m
at
r
ix
a
n
a
ly
s
is
an
d
R
OC
cu
r
v
es
a
r
e
p
lo
tted
to
v
is
u
alize
class
if
icatio
n
q
u
ality
.
T
r
ain
in
g
lo
s
s
an
d
ac
cu
r
ac
y
ar
e
m
o
n
ito
r
ed
o
v
er
ep
o
ch
s
to
d
et
ec
t
u
n
d
er
f
itti
n
g
o
r
o
v
er
f
itti
n
g
.
3
.
5
.
Co
m
pa
ra
t
iv
e
a
na
ly
s
is
T
o
v
alid
ate
th
e
ef
f
ec
tiv
en
es
s
o
f
th
e
p
r
o
p
o
s
ed
DM
O
-
C
NN
-
L
STM
f
r
am
ewo
r
k
,
we
co
n
d
u
cte
d
ex
p
er
im
en
ts
co
m
p
ar
in
g
its
p
er
f
o
r
m
an
ce
ag
ain
s
t
s
ev
er
al
tr
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
m
o
d
els
—
lo
g
is
tic
r
eg
r
ess
io
n
,
r
a
n
d
o
m
f
o
r
est
,
XG
B
o
o
s
t
—
an
d
tr
ad
itio
n
al
d
ee
p
le
ar
n
in
g
m
o
d
els
i
n
clu
d
in
g
ML
P,
C
NN,
an
d
L
STM
.
T
h
e
s
am
e
p
r
ep
r
o
ce
s
s
ed
d
ata
s
et
was
u
s
ed
ac
r
o
s
s
all
m
o
d
els
to
en
s
u
r
e
f
air
n
ess
.
T
h
e
DM
O
-
C
NN
-
L
ST
M
co
n
s
is
ten
tly
ac
h
iev
ed
s
u
p
e
r
io
r
r
esu
lts
in
all
ev
alu
atio
n
m
etr
ic
s
,
co
n
f
ir
m
in
g
its
r
o
b
u
s
tn
ess
an
d
p
r
ed
ictiv
e
p
o
wer
.
4.
RE
SU
L
T
S AN
D
AN
AL
Y
SI
S
T
h
e
p
r
o
p
o
s
ed
DM
O
-
C
NN
-
L
STM
m
o
d
el
was
ev
alu
ated
o
n
th
e
Diab
etes
1
3
0
-
US
Ho
s
p
itals
d
ataset,
co
n
tain
in
g
o
v
e
r
1
0
0
,
0
0
0
r
ec
o
r
d
s
an
d
5
5
clin
ical
f
ea
tu
r
es.
Af
t
er
p
r
e
p
r
o
ce
s
s
in
g
an
d
o
p
tim
izatio
n
,
th
e
m
o
d
el
was
co
m
p
ar
ed
ag
ain
s
t
s
ev
er
al
t
r
a
d
itio
n
al
m
ac
h
in
e
lear
n
in
g
m
o
d
e
ls
,
s
tan
d
alo
n
e
d
ee
p
lear
n
in
g
ar
ch
itectu
r
es,
a
n
d
s
tate
-
of
-
th
e
-
ar
t
h
y
b
r
id
class
if
ier
s
.
T
h
e
e
x
p
er
im
e
n
ts
aim
ed
t
o
m
ea
s
u
r
e
class
if
icatio
n
p
e
r
f
o
r
m
an
ce
u
s
in
g
v
a
r
io
u
s
s
tatis
t
ical
an
d
d
iag
n
o
s
tic
m
etr
ics
in
clu
d
in
g
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
,
an
d
ar
ea
u
n
d
e
r
th
e
R
OC
cu
r
v
e
(
AUC
-
R
O
C
)
.
4
.
1
.
E
x
perim
ent
a
l
c
o
nfig
ura
t
io
n
Fiv
e
ex
p
er
im
en
ts
wer
e
co
n
d
u
cted
with
v
ar
y
in
g
DM
O
p
o
p
u
latio
n
s
izes,
iter
atio
n
s
,
an
d
C
NN
-
L
STM
co
n
f
ig
u
r
atio
n
s
.
T
h
e
p
ar
a
m
eter
s
ettin
g
s
u
s
ed
f
o
r
th
ese
ex
p
er
im
en
ts
ar
e
p
r
esen
ted
in
T
ab
le
1
,
wh
ich
o
u
tlin
es
th
e
d
esig
n
o
f
th
e
DM
O
-
C
NN
-
L
STM
m
o
d
el.
T
h
ese
ex
p
e
r
im
en
t
s
wer
e
s
p
ec
if
ically
aim
e
d
at
e
v
alu
atin
g
t
h
e
im
p
ac
t
o
f
d
if
f
er
en
t
p
ar
am
eter
c
h
o
ices
o
n
ac
cu
r
ac
y
,
s
tab
ilit
y
,
an
d
c
o
n
v
er
g
en
ce
b
eh
a
v
io
r
.
T
h
e
o
u
tco
m
es,
wh
ich
h
ig
h
lig
h
t
th
e
r
elativ
e
ef
f
ec
tiv
e
n
ess
o
f
ea
ch
co
n
f
i
g
u
r
atio
n
,
ar
e
s
u
m
m
ar
i
ze
d
in
T
ab
les 2
a
n
d
3
.
Fig
u
r
e
3
ac
cu
r
ac
y
s
co
r
es
ac
r
o
s
s
5
ex
p
er
im
en
ts
o
f
DM
O
an
d
C
NN
-
L
STM
s
h
o
ws
a
co
n
s
is
ten
t
im
p
r
o
v
em
e
n
t
in
th
e
m
o
d
el'
s
class
if
icatio
n
ac
cu
r
ac
y
.
T
h
e
ac
cu
r
ac
y
p
er
ce
n
ta
g
e
s
tead
ily
in
cr
ea
s
es
f
r
o
m
ap
p
r
o
x
im
ately
9
1
.
2
%
i
n
E
x
p
er
im
en
t
1
to
its
h
ig
h
est
p
o
in
t
o
f
9
6
.
0
%
in
e
x
p
er
im
e
n
t
5
.
B
ased
o
n
th
e
f
in
al
ex
p
er
im
en
t
(
E
x
p
_
5
)
o
f
t
h
e
DM
O
-
C
NN
-
L
STM
m
o
d
el,
wh
ich
ac
h
iev
ed
th
e
h
ig
h
est
ac
cu
r
ac
y
o
f
9
6
.
1
%,
th
e
DM
O
alg
o
r
ith
m
s
elec
ted
th
e
f
o
llo
win
g
k
e
y
f
ea
t
u
r
es
f
r
o
m
th
e
o
r
ig
i
n
al
5
5
-
f
ea
tu
r
e
d
ataset.
T
h
ese
f
ea
t
u
r
es
ar
e
co
n
s
id
er
ed
th
e
m
o
s
t in
f
o
r
m
ativ
e
an
d
r
elev
an
t f
o
r
p
r
ed
ictin
g
d
iab
etes
, s
ee
T
ab
le
4
.
1
D
c
o
n
v
o
l
u
t
i
o
n
a
l
l
a
y
e
r
I
n
p
u
t
(
f
u
l
l
f
e
a
t
u
r
e
s
e
t
D
M
O
S
e
a
r
c
h
O
p
tim
iz
e
d
F
e
a
tu
re
su
b
se
t
M
a
x
P
o
o
l
i
n
g
La
y
e
r
O
u
t
p
u
t
La
y
e
r
D
e
n
se
La
y
e
r
si
g
m
o
i
d
a
c
t
i
v
a
t
i
o
n
Hyb
r
i
d
CN
N
-
L
S
T
M
M
e
mor
y
C
e
ll
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
6
,
Decem
b
e
r
20
25
:
5
5
5
5
-
5
5
6
9
5562
DM
O
ef
f
ec
tiv
ely
f
ilter
ed
1
5
k
ey
f
ea
tu
r
es
o
u
t
o
f
5
5
to
tal.
T
h
ese
f
ea
tu
r
es
s
p
an
d
em
o
g
r
ap
h
ics,
en
co
u
n
ter
h
is
to
r
y
,
m
ed
ical
d
iag
n
o
s
is
,
an
d
m
ed
icatio
n
u
s
ag
e,
wh
ich
ar
e
all
h
ig
h
ly
co
r
r
elate
d
with
d
ia
b
etes
r
is
k
.
Featu
r
es
lik
e
A1
C
r
esu
lt,
n
u
mb
er_
in
p
a
tien
t
,
n
u
m_
med
ic
a
tio
n
s
,
an
d
d
i
a
b
etesMed
ar
e
esp
ec
ially
im
p
ac
tf
u
l
in
p
r
e
d
ictin
g
d
iab
etic
s
tatu
s
.
T
ab
le
5
s
h
o
ws
th
e
im
p
o
r
ta
n
ce
s
co
r
es
o
f
th
e
s
elec
ted
f
ea
tu
r
es
b
y
th
e
DM
O
alg
o
r
ith
m
th
at
c
o
n
tr
ib
u
ted
to
th
e
h
ig
h
est
ac
cu
r
ac
y
(
9
6
.
1
%
)
in
p
r
ed
ictin
g
d
iab
etes
u
s
in
g
t
h
e
C
NN
-
L
STM
m
o
d
el.
T
h
ese
f
ea
tu
r
e
im
p
o
r
ta
n
ce
s
co
r
es
h
ig
h
lig
h
t
wh
ich
clin
ical
an
d
b
eh
a
v
io
r
al
v
ar
i
ab
les
p
la
y
ed
th
e
m
o
s
t
s
ig
n
if
ica
n
t
r
o
le
i
n
im
p
r
o
v
in
g
m
o
d
el
p
er
f
o
r
m
an
ce
.
B
y
id
en
tif
y
in
g
an
d
p
r
io
r
itizin
g
th
ese
f
ea
tu
r
es,
th
e
r
esu
lts
d
e
m
o
n
s
tr
ate
th
e
ef
f
ec
tiv
en
ess
o
f
t
h
e
DM
O
-
b
ased
s
elec
tio
n
p
r
o
ce
s
s
in
en
h
an
ci
n
g
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
m
o
d
el.
T
ab
le
1
.
Par
am
eter
s
ettin
g
s
f
o
r
DM
O
-
C
NN
-
L
STM
m
o
d
el
P
a
r
a
me
t
e
r
V
a
l
u
e
r
a
n
g
e
F
i
n
a
l
v
a
l
u
e
u
se
d
P
o
p
u
l
a
t
i
o
n
S
i
z
e
(
D
M
O
)
20
–
1
0
0
1
0
0
M
a
x
I
t
e
r
a
t
i
o
n
s
(
D
M
O
)
10
–
1
0
0
1
0
0
F
e
a
t
u
r
e
S
u
b
se
t
S
i
z
e
A
u
t
o
-
sel
e
c
t
e
d
b
y
D
M
O
-
C
N
N
La
y
e
r
s
2
2
C
N
N
F
i
l
t
e
r
S
i
z
e
3
×
3
3
×
3
P
o
o
l
i
n
g
T
y
p
e
M
a
x
P
o
o
l
i
n
g
M
a
x
P
o
o
l
i
n
g
LSTM
U
n
i
t
s
64
–
1
2
8
64
A
c
t
i
v
a
t
i
o
n
F
u
n
c
t
i
o
n
R
e
LU
+
S
i
g
m
o
i
d
R
e
LU
/
S
i
g
mo
i
d
O
p
t
i
mi
z
e
r
A
d
a
m
A
d
a
m
Ep
o
c
h
s
20
–
1
0
0
50
B
a
t
c
h
S
i
z
e
32
32
Lo
ss F
u
n
c
t
i
o
n
B
i
n
a
r
y
C
r
o
sse
n
t
r
o
p
y
B
i
n
a
r
y
C
r
o
sse
n
t
r
o
p
y
T
ab
le
2
.
E
x
p
er
im
en
tal
r
esu
lt
s
o
n
DM
O
-
C
NN
-
L
STM
(
v
alid
atio
n
ac
cu
r
ac
y
)
Ex
p
e
r
i
m
e
n
t
A
c
c
u
r
a
c
y
(
%)
M
S
E
(
%)
M
S
P
E
(
%)
D
M
O
T
i
me
(
ms)
C
N
N
-
LSTM
Ti
m
e
(
ms)
Ex
p
_
1
9
1
.
2
0
.
2
3
1
2
1
.
0
2
5
8
.
4
1
0
.
3
Ex
p
_
2
9
2
.
8
0
.
1
8
5
4
0
.
9
1
1
0
2
.
3
1
0
.
5
Ex
p
_
3
9
3
.
5
0
.
1
4
1
3
0
.
7
6
1
2
5
.
8
1
1
.
1
Ex
p
_
4
9
4
.
6
0
.
1
0
8
6
0
.
6
3
1
4
2
.
0
1
1
.
3
Ex
p
_
5
9
6
.
1
0
.
0
9
4
1
0
.
5
9
2
5
7
.
1
1
1
.
5
T
ab
le
3
.
E
x
p
er
im
en
tal
R
esu
lts
o
n
DM
O
-
C
NN
-
L
STM
(
v
alid
atio
n
ac
cu
r
ac
y
)
Ex
p
e
r
i
m
e
n
t
P
o
p
S
i
z
e
I
t
e
r
a
t
i
o
n
s
Ep
o
c
h
s
A
c
c
u
r
a
c
y
(
%)
Ex
p
_
1
20
5
20
9
1
.
2
Ex
p
_
2
40
10
20
9
2
.
8
Ex
p
_
3
60
40
40
9
3
.
5
Ex
p
_
4
80
40
50
9
4
.
6
Ex
p
_
5
1
0
0
1
0
0
50
9
6
.
1
Fig
u
r
e
3
.
Acc
u
r
ac
y
s
co
r
es a
cr
o
s
s
5
ex
p
er
im
en
ts
o
f
DM
O
an
d
C
NN
-
L
STM
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
DMO
-
CNN
-
LS
T
M fr
a
mewo
r
k
fo
r
fea
tu
r
e
s
elec
tio
n
a
n
d
d
i
a
b
etes
…
(
Mu
ta
s
em
K
.
A
ls
ma
d
i
)
5563
T
ab
le
4
.
Selecte
d
f
ea
tu
r
e
s
b
y
DM
O
(
E
x
p
_
5
)
f
o
r
o
p
tim
al
p
r
e
d
ictio
n
ac
cu
r
ac
y
N
o
.
F
e
a
t
u
r
e
N
a
me
D
e
scri
p
t
i
o
n
1
a
g
e
P
a
t
i
e
n
t
's
a
g
e
g
r
o
u
p
2
r
a
c
e
P
a
t
i
e
n
t
's ra
c
e
3
g
e
n
d
e
r
P
a
t
i
e
n
t
's
g
e
n
d
e
r
4
t
i
m
e
_
i
n
_
h
o
s
p
i
t
a
l
N
u
mb
e
r
o
f
d
a
y
s
a
d
mi
t
t
e
d
5
n
u
m
_
l
a
b
_
p
r
o
c
e
d
u
r
e
s
N
u
mb
e
r
o
f
l
a
b
t
e
s
t
s
p
e
r
f
o
r
me
d
6
n
u
m
_
p
r
o
c
e
d
u
r
e
s
N
u
mb
e
r
o
f
p
r
o
c
e
d
u
r
e
s
(
o
t
h
e
r
t
h
a
n
l
a
b
)
7
n
u
m
_
m
e
d
i
c
a
t
i
o
n
s
C
o
u
n
t
o
f
d
i
s
t
i
n
c
t
m
e
d
i
c
a
t
i
o
n
s
p
r
e
s
c
r
i
b
e
d
8
n
u
m
b
e
r
_
o
u
t
p
a
t
i
e
n
t
O
u
t
p
a
t
i
e
n
t
v
i
s
i
t
s
9
n
u
m
b
e
r
_
e
merg
e
n
c
y
Emerg
e
n
c
y
v
i
si
t
s
10
n
u
m
b
e
r
_
i
n
p
a
t
i
e
n
t
I
n
p
a
t
i
e
n
t
v
i
si
t
s
11
d
i
a
g
_
1
P
r
i
mary
d
i
a
g
n
o
s
i
s c
o
d
e
12
A
1
C
r
e
s
u
l
t
M
o
s
t
r
e
c
e
n
t
A
1
C
t
e
s
t
r
e
s
u
l
t
13
c
h
a
n
g
e
W
h
e
t
h
e
r
me
d
i
c
a
t
i
o
n
s wer
e
c
h
a
n
g
e
d
14
d
i
a
b
e
t
e
sM
e
d
W
h
e
t
h
e
r
d
i
a
b
e
t
e
s
me
d
i
c
a
t
i
o
n
w
a
s
p
r
e
scri
b
e
d
15
r
e
a
d
m
i
t
t
e
d
W
h
e
t
h
e
r
t
h
e
p
a
t
i
e
n
t
w
a
s re
a
d
mi
t
t
e
d
T
ab
le
5
.
Featu
r
e
im
p
o
r
tan
ce
ta
b
le
(
r
an
k
e
d
b
y
DM
O
s
elec
tio
n
weig
h
t
)
R
a
n
k
F
e
a
t
u
r
e
I
mp
o
r
t
a
n
c
e
S
c
o
r
e
(
0
–
1)
1
A
1
C
r
e
s
u
l
t
0
.
7
8
2
d
i
a
b
e
t
e
sM
e
d
0
.
7
5
3
r
e
a
d
m
i
t
t
e
d
0
.
7
4
4
a
g
e
0
.
7
2
5
n
u
m
_
m
e
d
i
c
a
t
i
o
n
s
0
.
7
0
6
t
i
m
e
_
i
n
_
h
o
s
p
i
t
a
l
0
.
6
8
7
n
u
m
b
e
r
_
i
n
p
a
t
i
e
n
t
0
.
6
6
8
d
i
a
g
_
1
0
.
6
3
9
n
u
m
_
l
a
b
_
p
r
o
c
e
d
u
r
e
s
0
.
6
0
10
n
u
m
_
p
r
o
c
e
d
u
r
e
s
0
.
5
8
11
r
a
c
e
0
.
5
5
12
c
h
a
n
g
e
0
.
5
3
13
n
u
m
b
e
r
_
o
u
t
p
a
t
i
e
n
t
0
.
5
1
14
n
u
m
b
e
r
_
e
merg
e
n
c
y
0
.
4
8
15
g
e
n
d
e
r
0
.
4
5
4
.
2
.
P
er
f
o
r
m
a
nce
m
e
t
rics
T
ab
les
6
a
n
d
7
s
u
m
m
ar
izes
th
e
class
if
icatio
n
m
etr
ics
o
f
all
co
m
p
ar
ed
m
o
d
els.
Am
o
n
g
th
e
tr
ad
itio
n
al
ML
m
o
d
els,
XGBo
o
s
t
o
u
tp
er
f
o
r
m
ed
L
o
g
is
tic
R
eg
r
ess
io
n
an
d
R
an
d
o
m
Fo
r
est
with
an
ac
cu
r
ac
y
o
f
9
4
.
0
%
an
d
F1
-
s
co
r
e
o
f
9
3
.
9
%.
Am
o
n
g
s
tan
d
alo
n
e
d
ee
p
lear
n
in
g
ar
ch
ite
c
tu
r
es,
th
e
Mu
ltil
ay
er
Per
ce
p
tr
o
n
(
ML
P)
ac
h
iev
ed
9
4
.
1
%
ac
c
u
r
ac
y
.
Ho
wev
er
,
th
e
p
r
o
p
o
s
ed
DM
O
-
C
NN
-
L
STM
m
o
d
el
ac
h
iev
ed
th
e
b
est
p
e
r
f
o
r
m
a
n
ce
ac
r
o
s
s
all
ev
alu
atio
n
cr
iter
ia,
with
an
ac
cu
r
ac
y
o
f
9
6
.
1
%,
p
r
ec
is
io
n
o
f
9
5
.
1
%,
r
ec
all
o
f
9
4
.
0
%,
F1
-
s
co
r
e
o
f
9
4
.
6
%,
an
d
R
OC
-
AU
C
o
f
0
.
9
6
.
T
ab
le
6
.
C
r
o
s
s
-
v
alid
atio
n
p
er
f
o
r
m
an
ce
m
et
r
ics co
m
p
ar
is
o
n
M
o
d
e
l
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
S
c
o
r
e
R
O
C
-
AUC
M
S
E
M
A
E
Lo
g
i
s
t
i
c
R
e
g
r
e
ssi
o
n
9
1
.
3
%
9
0
.
4
%
9
2
.
1
%
9
1
.
2
%
0
.
8
8
0
.
1
0
5
0
.
1
6
2
D
e
c
i
s
i
o
n
Tr
e
e
9
3
.
5
9
4
.
0
9
2
.
3
9
3
.
1
0
.
8
6
0
.
1
8
5
0
.
1
1
2
N
a
i
v
e
B
a
y
e
s
9
2
.
8
9
1
.
5
9
3
.
7
9
2
.
6
0
.
8
2
0
.
1
9
2
0
.
1
1
8
S
V
M
9
0
.
2
8
9
.
7
9
1
.
5
9
0
.
6
0
.
9
0
0
.
1
4
3
0
.
0
8
6
R
a
n
d
o
m F
o
r
e
s
t
9
3
.
2
%
9
2
.
5
%
9
3
.
8
%
9
3
.
1
%
0
.
9
2
0
.
0
8
9
0
.
1
2
4
X
G
B
o
o
st
9
4
.
0
%
9
3
.
7
%
9
4
.
1
%
9
3
.
9
%
0
.
9
4
0
.
0
7
1
0
.
0
9
8
M
LP
9
4
.
1
%
9
4
.
2
%
9
3
.
5
%
9
3
.
8
%
0
.
9
3
0
.
0
6
8
0
.
0
9
3
D
M
O
-
C
N
N
-
LSTM
9
6
.
1
%
9
5
.
1
%
9
4
.
0
%
9
4
.
6
%
0
.
9
6
0
.
0
4
3
0
.
6
3
T
ab
le
7
.
Statis
tical
an
aly
s
is
co
m
p
ar
is
o
n
s
M
o
d
e
l
A
v
g
A
c
c
u
r
a
c
y
S
t
d
D
e
v
Tr
a
i
n
T
i
me
(
s)
p
-
v
a
l
u
e
(
v
s D
M
O
)
Lo
g
i
s
t
i
c
R
e
g
r
e
ssi
o
n
9
1
.
3
%
±
0
.
6
%
2
.
3
<
0
.
0
0
1
D
e
c
i
s
i
o
n
Tr
e
e
8
9
.
5
%
±
0
.
8
%
1
.
9
<
0
.
0
0
1
N
a
i
v
e
B
a
y
e
s
8
8
.
0
%
±
0
.
9
%
0
.
7
<
0
.
0
0
1
S
V
M
9
2
.
5
%
±
0
.
5
%
1
1
.
3
<
0
.
0
0
1
R
a
n
d
o
m F
o
r
e
s
t
9
3
.
2
%
±
0
.
4
%
4
.
5
<
0
.
0
0
1
X
G
B
o
o
st
9
4
.
0
%
±
0
.
3
%
6
.
1
<
0
.
0
0
1
M
LP (
D
e
e
p
N
N
)
9
4
.
1
%
±
0
.
3
%
8
.
2
<
0
.
0
0
1
D
M
O
-
C
N
N
-
LSTM
9
6
.
1
%
±
0
.
2
%
1
9
.
3
–
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
6
,
Decem
b
e
r
20
25
:
5
5
5
5
-
5
5
6
9
5564
Fro
m
T
ab
le
6
,
ex
p
er
im
en
tal
r
esu
lts
d
em
o
n
s
tr
ate
th
at
o
u
r
DM
OA
-
C
NN
-
L
STM
m
o
d
el
ac
h
iev
es
a
p
r
ed
ictio
n
ac
cu
r
ac
y
o
f
9
6
.
1
%,
s
ig
n
if
ican
tly
o
u
tp
er
f
o
r
m
in
g
ex
is
tin
g
m
eth
o
d
s
in
clu
d
in
g
ML
PNN
(
9
4
.
1
%),
d
ec
is
io
n
tr
ee
(
9
3
.
5
%),
an
d
SVM
(
9
0
.
1
6
%).
I
n
ad
d
itio
n
,
DM
O
an
d
C
NN
-
L
STM
h
as
th
e
lo
west
MA
E
(
0
.
0
4
3
)
an
d
lo
west
MSE
(
0
.
0
6
3
)
,
in
d
i
ca
tin
g
th
e
s
m
allest
p
r
ed
ictio
n
er
r
o
r
a
m
o
n
g
all
m
o
d
els.
T
r
ad
i
tio
n
al
m
o
d
els
lik
e
lo
g
is
tic
r
eg
r
ess
io
n
an
d
r
an
d
o
m
f
o
r
est
p
er
f
o
r
m
r
ea
s
o
n
a
b
ly
b
u
t
ca
n
n
o
t
m
atch
th
e
p
r
ec
is
io
n
o
f
d
ee
p
n
eu
r
al
ar
ch
itectu
r
es.
XGBo
o
s
t
also
s
h
o
ws
s
tr
o
n
g
g
en
er
aliza
tio
n
an
d
p
er
f
o
r
m
s
s
ec
o
n
d
-
b
est
in
ter
m
s
o
f
MSE
an
d
MA
E
.
Fro
m
T
ab
le
7
,
K
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
c
o
n
f
ir
m
s
m
o
d
el
s
t
ab
ilit
y
.
ANOV
A/t
-
test
s
s
h
o
w
DM
O
an
d
C
NN
-
L
STM
s
ig
n
if
ican
tly
o
u
tp
er
f
o
r
m
s
b
aselin
es
(
p
<
0
.
0
1
)
,
wh
e
r
e
if
p
-
v
alu
e
<
0
.
0
5
,
th
e
d
if
f
er
e
n
c
e
is
s
tatis
tica
ll
y
s
ig
n
if
ican
t.
No
tab
ly
,
DM
O
an
d
C
NN
-
L
STM
i
s
m
o
r
e
co
m
p
u
tatio
n
ally
ex
p
en
s
iv
e,
b
u
t
o
f
f
er
s
s
u
p
er
io
r
ac
cu
r
ac
y
an
d
g
en
er
aliza
tio
n
,
in
o
th
er
wo
r
d
s
,
it
is
s
lo
wer
b
u
t
g
ain
s
s
u
p
er
io
r
ac
cu
r
ac
y
a
n
d
r
eliab
ilit
y
)
.
Naiv
e
B
ay
es
p
er
f
o
r
m
s
th
e
wea
k
est
o
v
er
all,
with
h
ig
h
est
MSE
an
d
lo
west
R
OC
AUC.
Dec
is
io
n
T
r
ee
is
f
ast
an
d
in
ter
p
r
etab
le
b
u
t
p
r
o
n
e
to
o
v
er
f
itti
n
g
.
SVM
s
h
o
ws
s
o
lid
p
er
f
o
r
m
an
ce
b
u
t
lo
n
g
er
tr
ain
in
g
tim
e
an
d
h
ig
h
er
m
em
o
r
y
u
s
ag
e.
DM
O
an
d
C
NN
-
L
STM
ac
h
ie
v
es
s
u
p
er
io
r
ac
cu
r
ac
y
,
l
o
west
er
r
o
r
,
an
d
h
ig
h
est
R
OC
,
AUC
ac
r
o
s
s
all
m
etr
ics.
Statis
t
ical
test
s
(
ANOV
A
an
d
t
-
test
s
)
r
ev
ea
l
all
p
-
v
alu
es
<
0
.
0
0
1
,
in
d
icatin
g
t
h
at
th
e
im
p
r
o
v
e
m
en
ts
o
f
DM
O
an
d
C
NN
-
L
STM
o
v
er
b
aselin
e
m
o
d
els ar
e
s
tatis
tical
ly
s
ig
n
if
ican
t.
4
.
3
.
Co
nfusi
o
n
m
a
t
rix
a
nd
RO
C
a
na
ly
s
is
T
h
e
co
n
f
u
s
io
n
m
atr
ix
f
o
r
th
e
DM
O
-
C
NN
-
L
STM
m
o
d
el
in
Fig
u
r
e
4
s
h
o
ws
a
h
ig
h
n
u
m
b
er
o
f
tr
u
e
p
o
s
itiv
es
(
T
P
=
5
8
5
)
a
n
d
tr
u
e
n
eg
ativ
es
(
T
N
=
5
9
0
)
,
in
d
icatin
g
th
e
m
o
d
el’
s
s
tr
o
n
g
d
is
cr
i
m
in
ativ
e
p
o
wer
.
T
h
e
n
u
m
b
er
o
f
f
alse
n
eg
ativ
es
(
FN
=
4
0
)
an
d
f
alse
p
o
s
itiv
es
(
FP
=
3
5
)
is
r
elativ
ely
lo
w,
wh
ich
r
ef
lects
g
o
o
d
g
en
er
aliza
tio
n
o
n
u
n
s
ee
n
s
am
p
les.
T
h
is
in
d
icate
s
a
well
-
b
alan
ce
d
class
if
ier
(
DM
O
an
d
C
NN
-
L
STM
)
.
T
h
e
R
OC
cu
r
v
e
i
n
Fig
u
r
e
5
co
n
f
ir
m
s
th
is
,
as
th
e
cu
r
v
e
clo
s
ely
a
p
p
r
o
ac
h
es
th
e
to
p
-
lef
t
c
o
r
n
er
,
in
d
i
ca
tin
g
a
h
ig
h
tr
u
e
p
o
s
iti
v
e
r
ate
an
d
a
lo
w
f
alse
p
o
s
itiv
e
r
ate
ac
r
o
s
s
v
ar
io
u
s
th
r
esh
o
ld
s
.
T
h
e
AUC
s
co
r
e
o
f
0
.
9
6
r
ea
f
f
ir
m
s
th
e
m
o
d
el'
s
ex
ce
llen
t c
lass
if
icat
io
n
ca
p
ab
ilit
y
.
Fig
u
r
e
4
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
DM
O
an
d
C
NN
-
L
STM
m
o
d
el
Fig
u
r
e
5
.
R
OC
C
u
r
v
e
o
f
DM
O
an
d
C
NN
-
L
STM
m
o
d
el
Evaluation Warning : The document was created with Spire.PDF for Python.