I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
15
,
No
.
4
,
A
u
g
u
s
t
20
26
,
p
p
.
3
7
3
2
~
3
7
4
4
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
4
.
p
p
3
7
3
2
-
3
7
4
4
3732
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Enha
ncing
ea
rly
detec
tion o
f
a
utis
m spectr
um diso
r
der t
h
ro
ug
h
ensem
ble
-
ba
sed
ma
chine learning
cla
ss
ifiers
Sh
a
beena
L
y
la
t
h,
L
a
x
m
i B
.
Ra
na
na
v
a
re
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
a
n
d
En
g
i
n
e
e
r
i
n
g
,
R
EV
A
U
n
i
v
e
r
si
t
y
,
B
e
n
g
a
l
u
r
u
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
2
,
2
0
2
4
R
ev
is
ed
J
u
l
13
,
2
0
2
6
Acc
ep
ted
J
u
l
21
,
2
0
2
6
Au
ti
sm
sp
e
c
tru
m
d
iso
r
d
e
r
(ASD)
is
a
d
e
v
e
l
o
p
m
e
n
tal
d
isa
b
i
li
ty
c
h
a
ra
c
teriz
e
d
b
y
si
g
n
ifi
c
a
n
t
so
c
ial,
c
o
m
m
u
n
ica
ti
o
n
,
a
n
d
b
e
h
a
v
i
o
ra
l
c
h
a
ll
e
n
g
e
s.
M
a
c
h
in
e
lea
rn
in
g
is
a
p
ra
c
ti
c
a
l
a
p
p
r
o
a
c
h
fo
r
a
u
ti
sm
d
e
tec
ti
o
n
.
T
h
e
p
r
o
p
o
se
d
e
n
se
m
b
le
-
b
a
se
d
m
a
c
h
in
e
lea
rn
in
g
c
las
sifier
m
e
th
o
d
o
l
o
g
y
p
re
se
n
t
e
d
in
t
h
is
stu
d
y
se
e
k
s
to
re
v
o
lu
t
io
n
ize
th
e
d
iag
n
o
sis
o
f
ASD
b
y
h
a
rn
e
ss
in
g
th
e
c
o
ll
e
c
ti
v
e
p
o
we
r
o
f
v
a
ri
o
u
s
m
a
c
h
in
e
lea
rn
in
g
a
l
g
o
rit
h
m
s.
Th
is
e
n
se
m
b
le
a
p
p
ro
a
c
h
is
d
e
sig
n
e
d
t
o
e
n
h
a
n
c
e
d
iag
n
o
stic
p
re
c
isi
o
n
,
m
it
ig
a
te
th
e
su
b
jec
ti
v
i
ty
a
ss
o
c
iate
d
with
trad
it
io
n
a
l
d
iag
n
o
stic
m
e
th
o
d
s,
a
n
d
a
c
c
e
ler
a
te
th
e
d
e
tec
ti
o
n
p
ro
c
e
ss
.
Th
is
m
e
th
o
d
o
l
o
g
y
a
d
d
re
ss
e
s
th
e
u
rg
e
n
t
n
e
e
d
fo
r
e
a
rl
y
a
n
d
a
c
c
u
ra
te
ASD
id
e
n
t
ifi
c
a
ti
o
n
,
e
n
a
b
li
n
g
ti
m
e
ly
in
ter
v
e
n
ti
o
n
s.
L
e
v
e
ra
g
in
g
c
o
m
p
lex
d
a
ta
a
n
a
l
y
sis,
i
t
o
ffe
rs
d
e
e
p
e
r
d
iag
n
o
stic
in
si
g
h
ts,
f
a
c
il
it
a
ti
n
g
in
fo
rm
e
d
c
li
n
ica
l
d
e
c
isio
n
s a
n
d
a
d
v
a
n
c
i
n
g
ASD
re
se
a
rc
h
.
Th
e
m
e
t
h
o
d
o
l
o
g
y
'
s
a
c
c
e
s
sib
il
it
y
a
c
ro
ss
h
e
a
lt
h
c
a
re
se
tt
in
g
s
m
a
rk
s
a
sig
n
if
ica
n
t
ste
p
f
o
rwa
rd
in
m
a
k
in
g
e
a
rly
ASD
d
e
tec
ti
o
n
m
o
re
u
n
i
v
e
rsa
ll
y
a
v
a
il
a
b
le,
sh
o
wc
a
sin
g
th
e
tran
sfo
rm
a
ti
v
e
p
o
ten
ti
a
l
o
f
m
a
c
h
i
n
e
lea
rn
in
g
in
h
e
a
lt
h
c
a
re
.
In
d
e
p
l
o
y
i
n
g
t
h
e
“
e
n
se
m
b
le
-
b
a
se
d
m
a
c
h
in
e
lea
rn
in
g
c
las
sifier
”
fo
r
ASD
d
iag
n
o
sis,
t
h
is
stu
d
y
u
ti
li
z
e
s
a
n
e
x
ten
siv
e
d
a
tas
e
t
c
o
m
p
risin
g
b
e
h
a
v
io
ra
l
a
n
d
m
e
d
ica
l
p
ro
fi
les
fro
m
d
i
v
e
rse
d
e
m
o
g
ra
p
h
ics
,
i
n
c
l
u
d
i
n
g
to
d
d
lers
,
c
h
i
ld
re
n
,
a
d
o
les
c
e
n
ts,
a
n
d
a
d
u
lt
s
with
A
S
D.
Up
o
n
t
h
e
p
re
li
m
in
a
ry
a
n
a
ly
sis,
th
e
d
a
tas
e
t
e
n
a
b
les
th
e
m
e
th
o
d
o
lo
g
y
to
lea
rn
fro
m
a
wid
e
a
rra
y
o
f
ASD
m
a
n
ifes
tatio
n
s,
e
n
su
rin
g
it
s
ro
b
u
stn
e
ss
a
n
d
a
p
p
l
ica
b
il
it
y
a
c
ro
s
s d
iffere
n
t
a
g
e
g
ro
u
p
s a
n
d
se
v
e
rit
y
lev
e
ls.
K
ey
w
o
r
d
s
:
Au
tis
m
d
etec
tio
n
Au
tis
m
s
p
ec
tr
u
m
d
is
o
r
d
er
E
n
s
em
b
le
-
b
ased
m
ac
h
in
e
Hea
lth
ca
r
e
L
ea
r
n
in
g
class
if
ier
Ma
ch
in
e
lear
n
in
g
Me
th
o
d
o
lo
g
y
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
:
Sh
ab
ee
n
a
L
y
lath
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
,
R
E
VA
Un
iv
er
s
ity
B
en
g
alu
r
u
,
I
n
d
ia
E
m
ail:
s
h
ab
ee
n
al_
1
2
@
r
ed
if
f
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
A
n
eu
r
o
d
ev
elo
p
m
en
tal
illn
ess
k
n
o
wn
as
au
tis
m
s
p
ec
tr
u
m
d
is
o
r
d
er
(
ASD)
u
s
u
ally
f
i
r
s
t
ap
p
ea
r
s
in
ea
r
ly
in
f
a
n
cy
a
n
d
last
s
th
r
o
u
g
h
o
u
t
ad
u
lth
o
o
d
[
1
]
.
T
h
er
e
is
ev
id
en
ce
o
f
b
eh
av
i
o
r
al
a
n
d
co
m
m
u
n
icatio
n
d
y
s
f
u
n
ctio
n
.
T
h
e
W
o
r
ld
Hea
lt
h
Or
g
an
izatio
n
(
W
HO)
esti
m
ates
th
at
1
%
o
f
p
eo
p
le
wo
r
ld
wid
e
ar
e
th
o
u
g
h
t
to
h
av
e
ASD
[
2
]
.
Ar
o
u
n
d
7
5
m
illi
o
n
p
eo
p
le
g
l
o
b
ally
ar
e
th
o
u
g
h
t
to
b
e
im
p
ac
ted
b
y
ASD.
Acc
o
r
d
in
g
t
o
r
esear
ch
f
in
d
in
g
s
,
co
-
o
cc
u
r
r
in
g
in
tell
ec
tu
al
im
p
air
m
en
t
is
p
r
esen
t
in
ar
o
u
n
d
o
n
e
-
th
ir
d
o
f
p
eo
p
le
with
ASD.
T
h
e
co
-
o
cc
u
r
r
en
ce
o
f
th
ese
ill
n
ess
es
[
2
]
,
m
ig
h
t
m
ak
e
it
d
if
f
i
cu
lt
f
o
r
p
e
o
p
le
to
lea
r
n
n
ew
t
h
in
g
s
an
d
ca
r
r
y
o
u
t
d
aily
d
u
ties
.
I
t
is
p
r
esen
tly
a
p
p
ar
en
t
th
at
t
h
er
e
is
a
lack
o
f
v
alid
ated
th
er
a
p
y
s
tr
ateg
ies
to
p
r
o
p
er
ly
m
a
n
ag
e
ASD,
an
d
th
e
ex
ac
t
o
r
ig
in
o
f
ASD
is
s
till
m
o
s
tly
u
n
clea
r
[
3
]
.
ASD
is
a
c
o
m
m
o
n
wo
r
l
d
wid
e
h
ea
lth
c
o
n
ce
r
n
th
at
ca
n
im
p
ac
t
in
d
iv
id
u
als,
f
a
m
ilies
,
an
d
t
h
e
c
o
m
m
u
n
ity
as
a
wh
o
le.
R
ec
en
t
f
i
n
d
in
g
s
s
u
g
g
est
th
at
p
er
m
an
en
t
b
r
ain
in
ju
r
y
is
n
o
t
th
e
ca
u
s
e
o
f
th
e
ab
n
o
r
m
al
b
eh
av
io
r
s
s
ee
n
in
ch
ild
r
en
with
ASD.
Alter
n
ativ
ely
,
th
ese
b
eh
av
io
r
s
m
i
g
h
t b
e
e
x
p
lain
ed
as th
e
b
r
ain
'
s
in
itial r
ea
ctio
n
to
a
s
tr
ess
f
u
l e
v
en
t
[
4
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
ea
r
ly
d
etec
tio
n
o
f
a
u
tis
m
s
p
ec
tr
u
m
d
is
o
r
d
er th
r
o
u
g
h
en
s
emb
le
-
b
a
s
ed
…
(
S
h
a
b
ee
n
a
Lyla
th
)
3733
E
ar
ly
in
ter
v
en
tio
n
tech
n
iq
u
es
m
u
s
t
b
e
p
u
t
in
to
p
r
ac
tice
to
r
ed
u
ce
t
h
e
s
tr
ess
an
d
wo
r
r
y
th
at
f
am
ilies
o
f
a
u
tis
tic
ch
ild
r
en
f
r
eq
u
en
tl
y
ex
p
er
ien
ce
.
T
h
e
p
r
o
g
r
a
m
s
er
v
es
as
an
i
n
s
tr
u
ctio
n
al
t
o
o
l,
g
iv
i
n
g
k
id
s
th
e
f
u
n
d
am
e
n
tal
in
f
o
r
m
atio
n
an
d
ab
ilit
ies
r
eq
u
ir
ed
f
o
r
th
ei
r
en
tire
d
e
v
elo
p
m
en
t
an
d
s
u
cc
ess
.
Ag
e
-
r
elate
d
d
ec
r
ea
s
es
in
h
u
m
an
b
r
ain
p
last
icity
h
av
e
b
ee
n
s
h
o
wn
in
p
r
io
r
s
tu
d
ies.
T
h
e
b
r
ai
n
'
s
ab
ilit
y
to
ad
ap
t
an
d
ch
a
n
g
e
is
k
n
o
wn
as
n
eu
r
al
p
last
icity
.
Fo
r
au
tis
tic
ch
ild
r
en
,
ea
r
ly
in
ter
v
en
tio
n
is
ad
v
is
ed
to
im
p
r
o
v
e
lan
g
u
ag
e
an
d
co
g
n
itiv
e
s
k
ills
b
ef
o
r
e
b
eh
av
i
o
r
al
p
r
o
b
lem
s
ar
is
e
[
5
]
.
Fo
r
th
is
r
ea
s
o
n
,
it
is
cr
itical
to
d
iag
n
o
s
e
ASD
as
s
o
o
n
as
p
o
s
s
ib
le.
A
ty
p
ical
ap
p
r
o
ac
h
in
th
e
d
ia
g
n
o
s
is
o
f
ASD
is
th
e
u
s
e
o
f
m
an
u
al
o
b
s
er
v
atio
n
s
.
T
h
e
p
r
e
v
io
u
s
ly
m
en
tio
n
ed
ap
p
r
o
ac
h
ca
n
b
e
d
escr
ib
ed
as
a
la
b
o
r
io
u
s
an
d
tim
e
-
co
n
s
u
m
in
g
p
r
o
ce
d
u
r
e
t
h
at
co
u
ld
p
r
o
v
id
e
d
if
f
icu
lties
wh
en
p
u
t
i
n
to
p
r
ac
tice.
On
e
s
tan
d
ar
d
ized
test
th
a
t
u
s
es
a
p
ar
e
n
t
-
ce
n
tr
ic
m
eth
o
d
to
d
ia
g
n
o
s
e
a
u
tis
m
is
th
e
m
o
d
if
ied
ch
ec
k
lis
t
f
o
r
au
tis
m
in
to
d
d
ler
s
(
M
-
C
HA
T
)
.
T
h
e
test
u
s
u
ally
r
eq
u
ir
es
a
s
ig
n
if
ican
t
tim
e
in
v
estme
n
t,
s
o
m
etim
es
s
p
an
n
in
g
m
an
y
h
o
u
r
s
.
I
n
th
er
ap
e
u
tic
s
ettin
g
s
th
at
ar
e
r
eg
u
lated
an
d
o
v
er
s
ee
n
,
clin
ica
l
p
er
s
o
n
n
el
u
s
u
ally
p
r
o
v
id
e
t
h
is
.
tr
ea
tm
en
t
[
6
]
.
I
t
tak
es
a
s
o
p
h
is
ticated
au
to
m
ated
id
e
n
tific
atio
n
tech
n
o
l
o
g
y
to
im
p
r
o
v
e
t
h
e
p
r
o
ce
s
s
's u
s
ab
ilit
y
an
d
e
f
f
icac
y
.
C
h
ild
r
en
wh
o
ar
e
ty
p
ically
d
ev
elo
p
in
g
(
T
D
)
ca
n
b
e
d
iag
n
o
s
ed
with
ASD
u
s
in
g
b
eh
av
io
r
al
an
d
p
h
y
s
io
lo
g
ical
c
r
iter
ia
[
7
]
.
Pe
o
p
le
wh
o
h
av
e
b
ee
n
d
iag
n
o
s
ed
with
ASD
f
in
d
it
d
if
f
icu
l
t
to
s
o
cialize
.
T
h
e
d
if
f
icu
lties
in
clu
d
e
a
r
an
g
e
o
f
n
o
n
v
er
b
al
elem
en
ts
,
in
clu
d
in
g
m
ain
tain
in
g
ey
e
c
o
n
tact
an
d
m
im
ick
in
g
f
ac
ial
ex
p
r
ess
io
n
s
.
Peo
p
le
c
o
u
ld
f
i
n
d
it
d
if
f
icu
lt
to
s
tay
f
o
cu
s
ed
,
en
g
ag
e
i
n
s
o
cial
s
itu
atio
n
s
,
an
d
ap
p
r
o
p
r
iately
ex
p
r
ess
th
eir
f
ee
lin
g
s
.
T
h
e
r
e
h
as
b
ee
n
a
lo
t
o
f
i
n
ter
est
in
th
e
r
esear
ch
o
n
e
y
e
g
az
e
f
ix
atio
n
in
k
id
s
with
ASD
.
E
lectr
o
en
ce
p
h
al
o
g
r
a
p
h
y
(
E
E
G)
an
d
m
o
b
ilit
y
test
s
wer
e
em
p
lo
y
ed
b
y
t
h
e
r
esear
ch
er
s
t
o
g
ath
er
i
n
f
o
r
m
atio
n
.
T
h
e
s
tu
d
y
'
s
co
n
clu
s
io
n
s
s
u
g
g
est
th
at
th
e
u
s
e
o
f
m
ac
h
in
e
lear
n
in
g
in
co
n
ju
n
ctio
n
with
ey
e
t
r
ac
k
in
g
tech
n
o
lo
g
y
m
ay
b
e
ab
le
to
d
is
tin
g
u
is
h
b
etwe
en
v
ar
io
u
s
v
is
u
al
tr
aits
in
p
eo
p
le
with
ASD
.
T
h
e
m
eth
o
d
o
u
tlin
ed
in
r
ef
er
en
ce
[
8
]
,
[
9
]
c
o
u
ld
h
elp
d
iag
n
o
s
e
au
tis
m
.
T
h
e
r
e
is
a
lo
t
o
f
p
r
o
m
is
e
in
u
s
in
g
ey
e
tr
ac
k
in
g
an
d
m
ac
h
i
n
e
lear
n
in
g
to
s
tu
d
y
ASD
.
B
ec
au
s
e
E
E
G
d
ata
ar
e
ess
en
tial
m
ar
k
er
s
o
f
a
b
er
r
a
n
t
b
r
ai
n
elec
tr
ic
al
ac
tiv
ity
lin
k
ed
to
n
eu
r
o
lo
g
ical
d
is
ea
s
es,
f
in
d
in
g
p
atter
n
s
in
th
e
d
ata
h
as a
ttra
cted
a
lo
t o
f
atten
tio
n
.
T
h
is
s
tu
d
y
aim
s
to
p
r
o
v
id
e
an
au
to
m
ated
ap
p
r
o
ac
h
f
o
r
th
e
d
i
ag
n
o
s
is
o
f
ASD
in
ch
ild
r
en
wh
o
ar
e
T
D.
T
h
e
p
r
o
ce
s
s
m
ak
es
u
s
e
o
f
a
q
u
an
titativ
e
a
p
p
r
o
ac
h
.
T
h
is
s
t
u
d
y
'
s
m
eth
o
d
o
lo
g
y
m
a
k
es
u
s
e
o
f
E
E
G
d
ata
a
n
d
co
n
ce
n
tr
ates
o
n
a
n
aly
zin
g
t
h
e
ar
ea
u
n
d
e
r
n
ea
th
a
s
ec
o
n
d
-
o
r
d
er
d
if
f
er
e
n
ce
p
lo
t
as
a
f
ac
to
r
t
h
at
s
ets
its
e
lf
ap
ar
t.
No
n
lin
ea
r
f
ea
tu
r
es
tak
en
f
r
o
m
E
E
G
d
ata
wer
e
u
s
ed
in
th
e
r
esear
ch
b
y
t
h
e
au
th
o
r
s
in
[
1
0
]
to
ass
ess
p
eo
p
le
with
ASD
.
T
h
e
s
ev
en
ty
-
th
r
e
e
EEG
m
ea
s
u
r
em
en
ts
f
r
o
m
s
ev
er
al
p
ar
ticip
an
ts
m
a
k
e
u
p
th
e
d
ataset.
Of
th
e
p
ar
ticip
an
ts
,
f
o
r
ty
-
o
n
e
s
h
o
wed
s
ig
n
s
o
f
ASD
,
wh
er
ea
s
th
e
o
t
h
er
th
ir
ty
-
two
s
h
o
wed
n
o
r
m
al
b
r
ain
ac
tiv
ity
.
T
h
e
E
E
G
d
ata
f
r
o
m
p
eo
p
le
with
ASD
wer
e
ev
alu
ated
b
y
th
e
r
esear
ch
er
s
u
s
in
g
th
e
Hig
u
ch
i
f
r
ac
tal
d
im
en
s
io
n
ap
p
r
o
ac
h
.
R
esear
ch
er
s
co
n
d
u
cted
a
s
tu
d
y
th
at
p
r
o
v
ed
th
e
ef
f
icac
y
o
f
th
is
tech
n
iq
u
e
in
ev
alu
atin
g
th
e
n
o
n
lin
ea
r
ity
o
f
E
E
G
s
ig
n
als
[
1
1
]
.
Acc
o
r
d
in
g
to
Allis
o
n
et
a
l.
[
1
2
]
,
p
eo
p
le
with
A
SD
h
ad
th
eir
E
E
G
b
r
ain
wav
es
an
aly
ze
d
to
e
x
tr
a
ct
d
if
f
er
en
t
f
ea
tu
r
es.
Au
to
r
eg
r
ess
iv
e
co
ef
f
icien
ts
,
m
u
ltifr
ac
tal
wav
elet
lead
er
esti
m
ates,
d
is
cr
ete
Fo
u
r
ier
tr
an
s
f
o
r
m
co
e
f
f
icien
ts
,
an
d
Sh
an
n
o
n
e
n
tr
o
p
y
wer
e
am
o
n
g
th
e
attr
ib
u
tes.
Par
ticip
an
ts
in
th
e
s
tu
d
y
wer
e
d
iv
id
ed
in
to
two
g
r
o
u
p
s
:
th
o
s
e
wh
o
h
ad
b
ee
n
o
f
f
icially
d
i
ag
n
o
s
ed
with
ASD
an
d
th
o
s
e
wh
o
d
id
n
o
t
d
is
p
lay
an
y
s
p
ec
if
ied
m
ed
ical
is
s
u
es
(
co
n
tr
o
l
s
u
b
jects).
T
o
ca
r
r
y
o
u
t
th
e
class
if
icatio
n
p
r
o
ce
s
s
,
m
ac
h
in
e
lear
n
in
g
m
o
d
els
wer
e
u
s
ed
.
T
o
im
p
r
o
v
e
E
E
G
s
ig
n
als,
a
d
ee
p
n
eu
r
al
n
et
wo
r
k
(
DNN)
m
o
d
el
ca
lled
th
e
two
-
d
im
en
s
io
n
al
-
d
e
ep
co
n
v
o
lu
tio
n
n
eu
r
al
n
etwo
r
k
(
2D
-
DC
NN
)
was u
s
ed
in
[
1
3
]
.
So
cial
s
k
ill
s
d
ev
elo
p
m
en
t
is
o
f
ten
d
if
f
icu
lt
f
o
r
ch
ild
r
en
w
ith
ASD
,
e
s
p
ec
ially
wh
en
it
co
m
es
to
n
o
n
v
e
r
b
al
ar
ea
s
lik
e
m
ain
tain
i
n
g
ey
e
c
o
n
tact
an
d
m
im
ick
in
g
f
ac
ial
ex
p
r
ess
io
n
s
.
Acc
o
r
d
in
g
to
Ali
et
a
l.
[
1
4
]
,
p
eo
p
le
with
ASD
d
o
n
o
t
s
u
cc
ee
d
ac
ad
em
ically
to
th
e
s
am
e
ex
ten
t
as
th
eir
p
ee
r
s
with
o
u
t
th
e
d
is
ea
s
e.
E
v
er
y
p
e
r
s
o
n
h
as
an
ex
p
r
ess
r
ig
h
t
to
ed
u
ca
tio
n
,
as
s
tated
in
ar
ticle
2
6
o
f
th
e
Un
iv
e
r
s
al
Dec
lar
atio
n
o
f
Hu
m
a
n
R
ig
h
ts
[
1
5
]
.
E
v
er
y
s
tu
d
en
t
h
a
s
th
e
r
ig
h
t
to
a
n
ed
u
ca
tio
n
,
r
e
g
ar
d
less
o
f
t
h
eir
f
i
n
an
cial
s
itu
atio
n
o
r
h
a
n
d
icap
.
Kid
s
with
ASD
ca
n
also
b
en
ef
it
f
r
o
m
th
e
p
r
ev
io
u
s
ly
d
escr
ib
ed
id
ea
.
E
d
u
ca
tin
g
ch
il
d
r
en
d
iag
n
o
s
ed
with
ASD
h
as
s
ev
er
al
d
if
f
ic
u
lties
.
Sp
ec
ial
ed
u
ca
tio
n
al
n
ee
d
(
SEN)
is
th
e
class
if
icatio
n
g
iv
e
n
to
k
i
d
s
wh
o
h
a
v
e
b
ee
n
id
en
tifie
d
as
n
ee
d
in
g
a
tailo
r
e
d
ed
u
ca
tio
n
al
a
p
p
r
o
ac
h
,
as
s
h
o
wn
in
[
1
6
]
.
E
n
co
u
r
ag
in
g
c
h
i
ld
r
en
with
SEN
to
g
et
s
p
ec
ialized
ed
u
ca
tio
n
ca
n
b
e
ch
allen
g
in
g
b
ec
au
s
e
o
f
s
o
cial
an
d
c
o
m
m
u
n
icatio
n
s
k
il
ls
is
s
u
es,
a
lack
o
f
p
ar
en
t
-
teac
h
er
co
llab
o
r
atio
n
,
an
d
lim
ited
r
eso
u
r
ce
s
an
d
s
u
p
p
o
r
t
o
p
tio
n
s
.
T
h
er
e
ar
e
s
ev
er
al
th
er
ap
e
u
tic
m
o
d
alities
av
ailab
le
to
d
iag
n
o
s
e
ASD
in
it
s
ea
r
ly
p
h
ases
.
W
h
en
ASD
is
v
er
y
lik
ely
to
o
cc
u
r
,
th
e
f
o
llo
win
g
d
iag
n
o
s
tic
m
eth
o
d
s
ar
e
f
r
e
q
u
e
n
tly
u
s
ed
in
clin
ical
s
ettin
g
s
.
T
h
is
r
esear
ch
is
d
r
iv
en
b
y
th
e
im
p
er
ativ
e
n
ee
d
to
im
p
r
o
v
e
A
SD
d
iag
n
o
s
is
th
r
o
u
g
h
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es,
aim
in
g
to
s
u
r
m
o
u
n
t
th
e
lim
itatio
n
s
o
f
tr
ad
itio
n
a
l
d
iag
n
o
s
tic
m
eth
o
d
s
th
at
ar
e
o
f
ten
s
u
b
jectiv
e
an
d
tim
e
-
co
n
s
u
m
in
g
.
L
ev
er
a
g
in
g
m
ac
h
in
e
lear
n
i
n
g
'
s
ab
ilit
y
to
d
is
ce
r
n
p
atter
n
s
in
e
x
ten
s
iv
e
d
atasets
,
th
is
s
tu
d
y
s
ee
k
s
to
d
ev
elo
p
o
b
jectiv
e,
ef
f
icien
t,
an
d
h
ig
h
l
y
ac
cu
r
ate
d
iag
n
o
s
tic
to
o
ls
f
o
r
ASD,
ad
d
r
ess
in
g
th
e
cr
itical
win
d
o
w
f
o
r
in
ter
v
e
n
tio
n
in
ea
r
ly
ch
ild
h
o
o
d
.
B
y
b
r
id
g
in
g
th
e
g
ap
b
etwe
en
ad
v
an
ce
d
co
m
p
u
tatio
n
al
m
eth
o
d
s
an
d
p
r
ac
tical
d
iag
n
o
s
tic
ap
p
licatio
n
s
,
th
e
r
esear
ch
en
d
ea
v
o
r
s
to
m
ak
e
ea
r
ly
ASD
d
etec
tio
n
m
o
r
e
ac
ce
s
s
ib
le
an
d
less
d
ep
en
d
en
t
o
n
s
p
ec
ialized
clin
ical
s
ettin
g
s
.
T
h
is
ef
f
o
r
t
n
o
t
o
n
ly
a
d
v
an
ce
s
o
u
r
s
cien
tific
u
n
d
er
s
tan
d
i
n
g
o
f
ASD
b
u
t
also
o
f
f
er
s
tan
g
i
b
le
s
o
lu
tio
n
s
to
en
h
an
ce
ea
r
l
y
in
ter
v
en
tio
n
s
tr
ateg
ies,
s
ig
n
if
ican
tly
im
p
ac
tin
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
7
3
2
-
3
7
4
4
3734
in
d
iv
id
u
als
with
ASD
an
d
th
eir
f
am
ilies
b
y
im
p
r
o
v
in
g
life
o
u
tco
m
es
an
d
in
teg
r
atin
g
tec
h
n
o
lo
g
y
i
n
s
o
lv
in
g
co
m
p
lex
h
ea
lth
ch
allen
g
es.
E
n
h
an
ce
d
d
iag
n
o
s
tic
ac
cu
r
a
cy
:
th
e
p
r
o
p
o
s
ed
“
en
s
em
b
l
e
-
b
ased
m
ac
h
in
e
lear
n
in
g
class
if
ier
”
in
tr
o
d
u
ce
s
a
cu
ttin
g
-
ed
g
e
d
iag
n
o
s
tic
to
o
l
f
o
r
ASD,
lev
e
r
ag
in
g
an
en
s
em
b
le
o
f
alg
o
r
ith
m
s
t
o
ac
h
ie
v
e
s
u
p
e
r
io
r
ac
cu
r
ac
y
a
n
d
a
d
ap
tab
ilit
y
in
d
etec
tin
g
th
e
wid
e
r
an
g
e
o
f
ASD
s
y
m
p
to
m
s
.
Ob
jectiv
e
a
n
d
e
f
f
icien
t
d
iag
n
o
s
is
:
th
is
m
eth
o
d
o
lo
g
y
s
tr
ea
m
lin
es
th
e
ASD
d
iag
n
o
s
is
p
r
o
ce
s
s
,
s
ig
n
if
ican
tly
r
ed
u
cin
g
r
eli
an
ce
o
n
s
u
b
jectiv
e
ass
es
s
m
en
ts
an
d
ex
p
ed
itin
g
d
iag
n
o
s
is
,
th
u
s
f
ac
ilit
atin
g
e
ar
lier
in
ter
v
e
n
tio
n
s
f
o
r
i
n
d
iv
id
u
als
with
ASD.
Data
-
d
r
iv
en
i
n
s
ig
h
ts
an
d
ac
c
ess
ib
ilit
y
:
b
y
h
ar
n
ess
in
g
co
m
p
lex
d
ata
p
atter
n
s
,
th
e
class
i
f
ier
p
r
o
v
id
es
d
ee
p
in
s
ig
h
ts
f
o
r
clin
ician
s
an
d
r
e
s
ea
r
ch
er
s
,
im
p
r
o
v
in
g
d
ec
is
io
n
-
m
ak
in
g
an
d
b
r
o
a
d
en
in
g
ac
ce
s
s
to
ea
r
ly
ASD
d
etec
tio
n
ac
r
o
s
s
d
iv
er
s
e
h
ea
lth
ca
r
e
en
v
ir
o
n
m
en
ts
.
T
h
e
r
esear
ch
is
o
r
g
an
ized
in
th
is
p
ap
er
in
5
s
ec
tio
n
s
.
Secti
o
n
1
is
th
e
in
tr
o
d
u
ctio
n
s
ec
tio
n
,
w
h
ich
g
iv
es
a
b
r
ief
o
v
e
r
v
iew
o
f
A
SD
.
S
ec
tio
n
2
g
iv
es
th
e
liter
atu
r
e
r
e
v
iew
.
S
ec
tio
n
3
d
esc
r
ib
es
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
l
o
g
y
.
S
ec
tio
n
4
g
iv
e
s
th
e
p
er
f
o
r
m
a
n
ce
e
v
alu
atio
n
o
f
th
e
r
esu
lt
in
t
h
e
f
o
r
m
o
f
g
r
ap
h
s
an
d
tab
les.
Fin
ally
,
s
ec
tio
n
5
co
n
cl
u
d
es b
y
s
u
m
m
ar
izin
g
th
e
m
ain
f
in
d
i
n
g
s
an
d
s
u
g
g
esti
n
g
f
u
t
u
r
e
r
ese
ar
ch
d
ir
ec
tio
n
s
.
2.
RE
L
AT
E
D
WO
RK
T
h
is
ASD
is
a
n
eu
r
o
d
e
v
elo
p
m
en
tal
d
is
ea
s
e
g
en
er
ally
d
etec
ted
th
r
o
u
g
h
o
u
t
ea
r
ly
in
f
an
cy
.
I
t
is
ch
ar
ac
ter
ized
b
y
p
r
o
b
lem
s
in
b
eh
av
io
r
an
d
co
g
n
itio
n
.
ASD
h
as
em
er
g
e
d
as
a
s
er
io
u
s
m
e
d
ical
co
n
ce
r
n
with
s
u
b
s
tan
tial
s
o
cieta
l
an
d
ec
o
n
o
m
ic
r
am
if
icatio
n
s
g
lo
b
all
y
.
B
ased
o
n
a
g
r
o
win
g
n
u
m
b
er
o
f
s
tu
d
ies,
it
h
as
b
ee
n
r
ev
ea
led
th
at
ab
er
r
an
t
b
e
h
av
io
r
in
ch
ild
r
en
with
ASD
m
ay
n
o
t
ex
clu
s
iv
ely
b
e
ca
u
s
ed
b
y
o
n
g
o
in
g
n
eu
r
o
lo
g
ical
ab
n
o
r
m
alities
,
b
u
t
also
b
y
t
h
e
ea
r
ly
ad
ap
tatio
n
o
f
th
e
b
r
ain
to
u
n
f
a
v
o
r
a
b
le
en
v
ir
o
n
m
en
tal
f
ac
to
r
s
[
1
7
]
.
C
h
ild
r
en
d
iag
n
o
s
ed
with
ASD
m
ay
en
co
u
n
ter
d
if
f
ic
u
lty
in
a
r
ea
s
s
u
ch
as
s
h
ar
ed
atten
tio
n
,
ey
e
co
n
tact,
s
o
cial
in
ter
ac
tio
n
s
,
an
d
em
o
tio
n
al
s
h
ar
in
g
.
T
h
e
u
s
er
h
as
s
u
p
p
lie
d
a
r
ef
er
e
n
ce
to
a
s
o
u
r
ce
,
r
ep
r
esen
ted
b
y
t
h
e
n
u
m
b
er
[
1
8
]
.
T
h
e
m
ac
h
in
e
le
ar
n
in
g
alg
o
r
ith
m
was
d
esig
n
e
d
to
d
iag
n
o
s
e
ch
ild
r
en
with
ASD
u
s
in
g
p
atter
n
s
r
etr
iev
ed
f
r
o
m
f
ac
ial
s
ca
n
s
.
T
h
e
m
o
d
el
attain
ed
a
n
ac
c
u
r
ac
y
o
f
8
2
.
5
1
%
in
its
ca
teg
o
r
i
za
tio
n
.
T
h
is
f
i
n
d
in
g
p
r
o
m
o
tes
th
e
in
v
esti
g
atio
n
o
f
m
ac
h
in
e
lear
n
in
g
alg
o
r
it
h
m
s
f
o
r
th
e
id
en
tific
atio
n
o
f
ASD
.
T
h
e
s
tu
d
y
s
u
cc
ess
f
u
lly
o
b
tain
ed
a
class
if
icatio
n
ac
cu
r
ac
y
o
f
8
4
%
u
tili
zin
g
a
m
ac
h
in
e
lear
n
in
g
ap
p
r
o
ac
h
to
d
is
cr
im
in
ate
b
etwe
en
th
e
ey
e
f
ix
atio
n
s
o
f
ch
ild
r
en
with
ASD
an
d
T
D
c
h
ild
r
en
.
T
h
e
f
o
llo
win
g
r
esear
ch
g
iv
es
ev
id
en
ce
o
f
th
e
ef
f
icac
y
an
d
im
p
a
r
tiality
b
en
ef
its
o
f
m
ac
h
in
e
lea
r
n
in
g
as
co
m
p
ar
ed
to
tr
ad
itio
n
al
d
iag
n
o
s
tic
s
ca
les.
Fu
r
th
er
m
o
r
e
,
as
co
m
p
ar
ed
to
T
D
y
o
u
n
g
s
ter
s
,
p
er
s
o
n
s
with
ASD
h
av
e
d
if
f
icu
lty
in
n
o
n
v
e
r
b
al
co
m
m
u
n
icatio
n
ab
ilit
ies,
s
u
ch
as e
x
p
r
ess
io
n
im
itatio
n
an
d
f
ac
ial
e
x
p
r
ess
io
n
r
ec
o
g
n
itio
n
(
FER).
T
h
e
s
tu
d
y
d
o
n
e
b
y
th
e
a
u
th
o
r
s
in
[
1
8
]
–
[
2
0
]
em
p
lo
y
ed
f
ac
ial
m
u
s
cle
an
aly
s
is
as
a
to
o
l
to
test
th
e
ca
p
ac
ity
o
f
ch
ild
r
en
with
ASD
to
co
p
y
th
e
f
ac
ial
ex
p
r
ess
io
n
s
o
f
o
th
er
s
.
T
h
e
f
in
d
in
g
s
s
h
o
w
th
at
th
e
r
ep
r
o
d
u
ctio
n
o
f
n
o
r
m
al
f
ac
ial
ex
p
r
ess
io
n
s
m
ig
h
t
s
er
v
e
as
a
b
eh
av
i
o
r
al
s
ig
n
al
f
o
r
th
e
d
ia
g
n
o
s
is
o
f
c
h
ild
r
en
with
ASD
.
A
tech
n
iq
u
e
was
cr
ea
ted
to
au
to
m
atica
lly
d
iag
n
o
s
e
ASD
in
p
er
s
o
n
s
wi
th
atten
tio
n
d
ef
icit
h
y
p
er
ac
tiv
ity
d
is
o
r
d
er
[
1
1
]
.
T
h
ey
b
u
ilt
a
b
r
an
d
-
n
ew
s
em
i
-
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
f
r
am
ewo
r
k
ca
lled
clu
s
ter
in
g
-
b
ased
au
tis
tic
tr
ait
class
if
icatio
n
(
C
AT
C
)
to
ac
co
m
p
lis
h
th
is
aim
.
T
h
r
o
u
g
h
th
e
ap
p
licatio
n
o
f
class
if
icatio
n
tech
n
iq
u
es,
C
A
T
C
ev
alu
ates
clas
s
if
ier
s
u
s
in
g
a
clu
s
ter
in
g
s
tr
ateg
y
.
I
n
s
tea
d
o
f
u
tili
zin
g
s
co
r
e
s
y
s
tem
s
lik
e
m
an
y
o
th
er
ASD
s
cr
ee
n
in
g
ap
p
r
o
ac
h
es,
th
is
m
eth
o
d
o
l
o
g
y
d
etec
ts
lik
ely
o
cc
u
r
r
en
ce
s
o
f
a
u
tis
m
b
ased
o
n
r
elate
d
tr
aits
.
T
h
e
em
p
ir
ical
f
in
d
i
n
g
s
,
wh
ic
h
ca
m
e
f
r
o
m
a
r
a
n
g
e
o
f
d
atas
ets
in
clu
d
in
g
k
id
s
,
teen
ag
er
s
,
an
d
a
d
u
lts
,
wer
e
v
alid
ated
an
d
c
o
n
tr
asted
with
o
th
er
wid
ely
-
u
s
ed
m
ac
h
in
e
lear
n
in
g
class
if
icatio
n
ap
p
r
o
ac
h
es.
T
h
e
r
esu
lts
d
em
o
n
s
tr
ated
th
at
wh
en
co
m
p
ar
ed
to
o
th
er
in
tellig
en
t
class
if
icatio
n
ap
p
r
o
ac
h
es
lik
e
ar
tific
ial
n
eu
r
al
n
etwo
r
k
s
(
A
NN)
,
r
an
d
o
m
f
o
r
est
(
R
F)
,
r
an
d
o
m
tr
ee
s
,
a
n
d
r
u
le
in
d
u
c
tio
n
,
C
AT
C
g
iv
es
class
if
ier
s
w
ith
g
r
ea
ter
p
r
o
jec
ted
ac
cu
r
ac
y
,
s
en
s
itiv
ity
,
an
d
s
p
ec
if
icity
.
E
s
ler
et
a
l.
[
1
6
]
em
p
lo
y
ed
m
ac
h
in
e
lear
n
in
g
al
g
o
r
ith
m
s
to
o
v
er
co
m
e
s
ig
n
if
ican
t
a
u
tis
m
-
r
elate
d
ch
allen
g
es.
T
o
in
c
r
ea
s
e
class
if
icatio
n
ac
cu
r
ac
y
,
th
ey
co
n
ce
n
t
r
ate
a
m
ajo
r
f
o
c
u
s
o
n
th
e
s
elec
tiv
e
s
elec
tio
n
o
f
im
p
o
r
tan
t
au
tis
m
tr
aits
.
T
h
ey
u
n
d
er
lin
e
h
o
w
v
ital
it
is
to
p
ick
th
e
r
elev
an
t
f
ea
tu
r
es
an
d
lim
it
d
ata
co
m
p
le
x
ity
to
g
en
er
ate
o
p
tim
is
tic
f
in
d
i
n
g
s
wh
en
d
iag
n
o
s
in
g
ASD.
T
h
e
au
th
o
r
s
also
n
o
te
a
v
ar
iety
o
f
f
ac
to
r
s
th
at
m
a
y
im
p
air
ac
cu
r
a
cy
,
s
u
ch
as
f
ea
tu
r
e
r
ed
u
n
d
an
c
y
,
in
ap
p
r
o
p
r
iate
s
am
p
lin
g
p
r
o
ce
d
u
r
es,
an
d
im
b
alan
ce
d
an
d
in
s
u
f
f
icien
t
s
am
p
le
s
izes.
C
o
u
p
led
RF
,
class
if
icatio
n
an
d
r
eg
r
ess
io
n
tr
ee
s
(
C
AR
T
)
an
d
RF
iter
ativ
e
d
ich
o
to
m
is
er
3
(
I
D3
)
,
ass
ess
in
g
th
e
tr
ain
ed
m
o
d
el
o
n
a
s
im
ilar
d
ataset
b
ef
o
r
e
d
ep
l
o
y
in
g
it
in
a
m
o
b
ile
ap
p
licatio
n
[
1
7
]
.
I
n
d
if
f
er
e
n
t
r
esear
ch
,
L
iu
et
a
l.
[
1
8
]
lo
o
k
ed
at
2
5
m
ac
h
in
e
lear
n
i
n
g
class
if
ier
s
u
s
in
g
a
g
ath
er
e
d
d
ataset
f
o
r
ASD
an
d
f
o
u
n
d
th
at
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
e
(
SVM
)
b
ased
o
n
s
eq
u
e
n
tial m
in
im
al
o
p
tim
izatio
n
(
SMO)
p
er
f
o
r
m
e
d
b
etter
in
th
eir
test
en
v
ir
o
n
m
e
n
t.
Acc
o
r
d
in
g
to
J
ian
g
et
a
l.
[
1
9
]
,
p
eo
p
le
with
ASD
h
av
e
s
ev
er
a
l
d
is
tin
ctiv
e
tr
aits
.
Ask
in
g
in
a
p
p
r
o
p
r
iate
q
u
esti
o
n
s
,
ac
tin
g
er
r
atica
lly
,
im
itatin
g
o
th
er
s
s
p
o
n
tan
eo
u
s
l
y
,
d
is
p
lay
in
g
a
lack
o
f
in
ter
e
s
t
in
o
th
er
p
eo
p
le,
f
in
d
in
g
it
d
if
f
ic
u
lt
to
s
h
ar
e
m
ea
ls
,
an
d
u
tili
zin
g
o
b
jects
r
ep
ea
ted
ly
ar
e
am
o
n
g
th
e
b
eh
av
i
o
r
s
th
at
ar
e
n
o
ted
.
Sam
ad
et
a
l.
[
2
0
]
l
o
o
k
e
d
at
war
n
in
g
in
d
icato
r
s
th
at
co
u
ld
p
o
in
t
to
th
e
ex
is
ten
ce
o
f
A
SD
in
th
eir
s
tu
d
y
.
Sp
ec
if
ic
in
d
icatio
n
s
o
f
AS
D
in
clu
d
ed
d
elay
e
d
s
p
ee
c
h
d
ev
elo
p
m
e
n
t,
r
e
p
etitiv
e
p
lay
h
a
b
its
,
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
ea
r
ly
d
etec
tio
n
o
f
a
u
tis
m
s
p
ec
tr
u
m
d
is
o
r
d
er th
r
o
u
g
h
en
s
emb
le
-
b
a
s
ed
…
(
S
h
a
b
ee
n
a
Lyla
th
)
3735
co
m
m
u
n
icatio
n
ch
allen
g
es.
I
t
was
also
n
o
ted
th
at
th
er
e
w
as
a
lack
o
f
f
ac
ial
ex
p
r
ess
io
n
,
p
r
eten
d
p
lay
,
an
d
im
ag
in
ativ
e
p
la
y
,
i
n
a
d
d
itio
n
to
a
p
r
o
p
en
s
ity
f
o
r
en
g
a
g
in
g
in
ac
tiv
e
p
ee
r
i
n
ter
ac
tio
n
,
u
n
d
er
s
tan
d
in
g
ir
o
n
y
,
an
d
m
ain
tain
in
g
p
e
r
s
o
n
al
s
p
ac
e.
Sam
ad
et
a
l.
[
2
0
]
s
h
o
w
t
h
at
th
eir
s
tu
d
y
in
cl
u
d
es
th
e
f
o
llo
win
g
f
ea
tu
r
es:
th
e
d
en
s
ity
o
f
q
u
ick
ey
e
m
o
v
em
e
n
ts
,
th
e
d
en
s
ity
o
f
m
u
s
cle
twitch
es,
an
d
th
e
f
ir
s
t
d
elay
o
f
r
a
p
id
ey
e
m
o
v
em
e
n
t
(
R
E
M)
.
I
t
h
as
b
ee
n
n
o
te
d
th
a
t
r
ep
etitiv
e
b
eh
av
io
r
s
,
s
u
ch
a
s
r
ep
ea
tin
g
m
o
to
r
a
n
d
s
en
s
o
r
y
ac
tio
n
s
,
ar
e
m
o
r
e
co
m
m
o
n
in
p
e
o
p
le
with
au
tis
m
,
esp
ec
ially
in
y
o
u
n
g
er
ag
e
g
r
o
u
p
s
.
T
h
ese
ac
tio
n
s
ar
en
'
t
s
ee
n
as
im
p
o
r
tan
t
s
ig
n
s
o
f
th
e
illn
ess
,
eith
er
.
On
th
e
o
th
er
h
an
d
,
o
ld
e
r
p
eo
p
le
with
h
ig
h
er
in
tellig
en
ce
q
u
o
ti
en
ts
(
I
Qs
)
ty
p
ically
d
is
p
lay
m
o
r
e
in
tr
icate
an
d
s
o
p
h
is
ticated
r
ec
u
r
r
en
t
b
e
h
av
io
r
al
p
atter
n
s
.
E
v
en
th
o
u
g
h
th
e
af
o
r
em
en
tio
n
ed
s
tu
d
y
s
h
o
wed
a
co
n
s
id
er
ab
le
in
c
r
ea
s
e
in
th
e
p
r
ed
ictio
n
o
f
ASD
,
th
e
ac
cu
r
ac
y
o
f
ac
ad
em
ic
p
e
r
f
o
r
m
an
ce
m
o
d
els
h
as
n
o
t y
et
ac
h
ie
v
ed
its
f
u
ll p
o
ten
t
ial.
R
u
le
-
b
ased
m
ac
h
in
e
le
ar
n
in
g
(
R
ML
)
ap
p
r
o
ac
h
es
wer
e
u
tili
ze
d
in
[
1
7
]
to
ex
am
in
e
th
e
ch
ar
ac
ter
is
tics
o
f
ASD
.
R
esear
ch
er
s
d
is
co
v
er
ed
th
at
ap
p
ly
i
n
g
R
ML
tech
n
iq
u
es
ca
n
g
r
ea
tly
in
cr
ea
s
e
cla
s
s
if
icatio
n
m
o
d
els
'
ac
cu
r
ac
y
in
d
etec
tin
g
in
s
tan
ce
s
o
f
ASD
.
I
n
a
wo
r
k
th
at
was
p
u
b
lis
h
ed
in
[
1
8
]
,
th
e
r
esear
ch
er
s
-
b
u
ilt
p
r
ed
ictio
n
m
o
d
els
f
o
r
p
e
o
p
le
o
f
v
ar
io
u
s
ag
es,
in
clu
d
in
g
ad
u
lts
,
ad
o
lescen
ts
,
an
d
ch
ild
r
en
,
u
s
in
g
th
e
R
F
an
d
I
D3
alg
o
r
ith
m
s
.
A
u
n
iq
u
e
ass
ess
m
en
t
to
o
l
th
at
co
m
b
in
es
th
e
au
tis
m
d
iag
n
o
s
tic
in
ter
v
iew
-
r
ev
is
ed
(
ADI
-
R
)
an
d
au
tis
m
d
iag
n
o
s
tic
o
b
s
er
v
ati
o
n
s
ch
ed
u
le
(
ADOS
)
m
ac
h
in
e
lear
n
in
g
a
p
p
r
o
ac
h
es
h
as
b
ee
n
p
r
o
p
o
s
ed
[
1
9
]
.
Div
er
s
e
attr
ib
u
te
en
c
o
d
in
g
m
eth
o
d
s
h
a
v
e
b
ee
n
u
tili
ze
d
to
tack
le
p
r
o
b
lem
s
ass
o
ciate
d
with
ir
r
eg
u
lar
d
ata
,
non
-
lin
ea
r
ity
,
an
d
in
s
u
f
f
icien
t
d
ata.
T
h
a
b
tah
et
a
l.
[
1
3
]
u
s
ed
co
g
n
itiv
e
co
m
p
u
tin
g
in
co
n
ju
n
ctio
n
with
d
ec
is
io
n
tr
ee
s
(
DT
)
,
lo
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
,
an
d
SVM
as
p
r
o
g
n
o
s
tic
an
d
d
iag
n
o
s
tic
class
if
ier
s
f
o
r
ASD.
E
x
am
in
in
g
th
e
c
o
r
r
elatio
n
v
alu
es
b
etwe
en
th
e
tr
aits
an
d
class
es
is
th
e
aim
o
f
th
is
in
v
esti
g
atio
n
.
T
h
e
cu
r
r
en
t
in
v
esti
g
atio
n
s
ee
k
s
to
f
u
r
t
h
er
e
ar
lier
s
tu
d
ies o
n
th
e
to
p
ic
[
1
7
]
.
Acc
o
r
d
in
g
t
o
Sam
ad
et
a
l.
[
2
0
]
,
th
e
r
e
wer
e
in
s
tan
ce
s
o
f
T
D
(
N
=
1
9
)
an
d
ASD
(
N
=
1
1
)
.
A
co
r
r
elatio
n
-
b
ased
attr
ib
u
te
s
elec
tio
n
m
eth
o
d
was
ap
p
lied
to
ascer
tain
th
e
s
ig
n
if
ican
ce
o
f
th
e
q
u
alities
.
A
s
in
g
le
d
ataset
wa
s
u
s
ed
f
o
r
th
e
m
o
d
el'
s
a
s
s
e
s
s
m
en
t,
an
d
th
e
co
m
p
ar
is
o
n
'
s
r
an
g
e
was
co
n
s
tr
ain
ed
.
Au
tis
m
is
ca
teg
o
r
ized
u
s
in
g
th
e
m
ac
h
in
e
lear
n
in
g
ap
p
r
o
ac
h
cr
ea
te
d
b
y
th
e
au
th
o
r
s
o
f
[
1
7
]
.
On
t
h
e
R
F
alg
o
r
ith
m
,
th
e
m
o
d
el
was
b
u
ilt.
T
h
e
lin
ea
r
d
is
cr
im
in
an
t
an
aly
s
is
(
L
DA
)
an
d
k
-
n
ea
r
est
n
eig
h
b
o
r
(
KN
N
)
alg
o
r
ith
m
s
wer
e
u
s
ed
in
[
1
6
]
t
o
d
iag
n
o
s
e
ASD
in
ch
ild
r
en
b
etwe
en
th
e
a
g
es
o
f
4
an
d
1
1
.
C
h
en
et
a
l.
[
1
7
]
p
u
t
o
u
t
a
p
ar
a
d
ig
m
f
o
r
ASD
in
2
0
1
8
.
C
h
ild
r
e
n
b
e
twee
n
th
e
ag
es
o
f
4
an
d
1
1
ar
e
th
e
m
o
d
el'
s
tar
g
et
m
ar
k
et.
T
h
e
R
F
clas
s
if
ier
i
s
em
p
lo
y
ed
.
L
iu
et
a
l.
[
1
8
]
lo
o
k
ed
at
h
o
w
well
th
e
DNN
co
u
ld
d
iag
n
o
s
e
ASD
.
T
wo
d
is
tin
ct
ad
u
lt
d
atasets
wer
e
u
s
ed
in
th
e
s
tu
d
y
.
L
iu
et
a
l.
[
1
8
]
cr
ea
ted
a
s
m
ar
tp
h
o
n
e
ap
p
licatio
n
p
r
o
g
r
am
m
in
g
in
ter
f
ac
e
(
API
)
in
2
0
1
9
th
a
t
m
ak
es
it
p
o
s
s
ib
le
to
d
iag
n
o
s
e
ASDs
in
p
eo
p
le
o
f
all
ag
es.
T
h
e
tech
n
o
l
o
g
ies
R
F
-
C
A
R
T
an
d
R
F
-
I
D3
s
u
p
p
o
r
te
d
th
e
in
ter
f
ac
e
.
J
ian
g
et
a
l.
[
1
9
]
lo
o
k
e
d
at
h
o
w
well
s
ev
er
al
SVM
k
er
n
els
r
ec
o
g
n
ized
d
ata
r
elate
d
to
ASD
in
y
o
u
n
g
s
ter
s
.
I
t w
as d
is
co
v
er
ed
th
at
th
e
p
o
ly
n
o
m
ial
k
e
r
n
el
p
e
r
f
o
r
m
ed
n
o
ticea
b
ly
b
etter
.
I
n
a
s
tu
d
y
th
at
was
p
u
b
lis
h
ed
in
[
1
1
]
,
th
e
r
esear
c
h
er
s
u
s
ed
f
o
u
r
d
atasets
th
at
wer
e
ass
o
c
iated
with
ASD
to
in
v
esti
g
ate
s
ev
er
al
f
ea
tu
r
e
s
elec
tio
n
tech
n
iq
u
es.
T
h
e
in
v
esti
g
atio
n
'
s
co
n
clu
s
io
n
s
s
h
o
wed
th
at
th
e
SVM
clas
s
if
ier
p
er
f
o
r
m
ed
b
etter
th
an
o
th
er
m
et
h
o
d
s
in
th
e
in
s
tan
ce
o
f
th
e
r
ep
ea
ted
in
c
r
em
en
tal
p
r
u
n
i
n
g
to
p
r
o
d
u
ce
er
r
o
r
r
ed
u
ctio
n
(
R
I
PP
E
R
)
-
b
ased
to
d
d
ler
s
u
b
g
r
o
u
p
.
W
h
en
ap
p
lied
to
th
e
ad
u
lt
s
u
b
s
et
u
s
in
g
th
e
co
r
r
elatio
n
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
(
C
FS
)
f
ea
tu
r
e
s
elec
tio
n
ap
p
r
o
ac
h
an
d
th
e
c
h
ild
s
u
b
s
et
u
s
in
g
th
e
C
FS
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
t
o
g
eth
er
with
th
e
B
o
r
u
ta
C
FS
in
ter
s
ec
t
(
B
I
C
)
m
eth
o
d
o
lo
g
y
,
th
e
SVM
class
if
ier
s
h
o
wed
b
etter
p
er
f
o
r
m
an
ce
.
Sev
e
r
al
f
ea
tu
r
e
s
u
b
s
ets
wer
e
an
aly
ze
d
b
y
t
h
e
r
e
s
ea
r
ch
er
s
u
s
in
g
th
e
Sh
ap
ley
ad
d
itiv
e
ex
p
lan
atio
n
s
(
SHAP)
ap
p
r
o
ac
h
.
T
h
is
s
tr
ateg
y
y
ield
ed
th
e
h
ig
h
est
lev
el
o
f
ac
cu
r
ac
y
a
f
ter
ev
alu
atin
g
ea
ch
f
ea
tu
r
e'
s
p
er
f
o
r
m
an
ce
.
T
o
ca
teg
o
r
ies
Par
k
i
n
s
o
n
'
s
illn
es
s
an
d
ASD
,
Sam
ad
et
a
l.
[
2
0
]
u
s
ed
en
s
em
b
le
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es
s
u
ch
as
f
u
zz
y
KNN
,
k
er
n
el
SVM
,
f
u
zz
y
co
n
v
o
lu
ti
o
n
n
eu
r
al
n
etwo
r
k
(
FC
NN)
,
an
d
R
F.
T
h
e
leav
e
-
one
-
p
er
s
o
n
-
o
u
t
cr
o
s
s
v
alid
at
io
n
(
L
OPOC
V)
m
eth
o
d
is
u
s
ed
to
v
alid
ate
th
e
ca
teg
o
r
izatio
n
f
i
n
d
in
g
s
.
3.
M
E
T
H
O
D
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
o
u
tlin
es
a
co
m
p
r
e
h
en
s
iv
e
ap
p
r
o
ac
h
f
o
r
d
e
v
elo
p
in
g
a
m
ac
h
in
e
lear
n
in
g
m
o
d
el
to
p
r
ed
ict
ASD
,
s
tar
tin
g
f
r
o
m
d
ata
u
n
d
e
r
s
tan
d
in
g
an
d
clea
n
in
g
,
tr
a
n
s
f
o
r
m
atio
n
,
f
e
atu
r
e
s
elec
tio
n
,
an
d
d
ata
p
r
o
ce
s
s
in
g
.
I
t
em
p
h
asiz
es
th
e
im
p
o
r
tan
ce
o
f
p
r
elim
i
n
ar
y
d
ata
a
n
aly
s
is
,
in
clu
d
in
g
clea
n
in
g
an
d
d
ata
p
r
ep
ar
atio
n
,
to
p
r
e
p
ar
e
th
e
d
ataset
f
o
r
m
o
d
ellin
g
.
A
n
en
s
em
b
le
class
if
ier
m
o
d
el,
in
co
r
p
o
r
atin
g
v
ar
i
o
u
s
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es
lik
e
ex
tr
a
tr
ee
s
(ET)
,
lig
h
t
g
r
ad
ien
t
-
b
o
o
s
tin
g
m
ac
h
in
e
(
L
ig
h
tGB
M
)
,
r
id
g
e
class
if
icatio
n
,
an
d
L
DA
,
is
p
r
o
p
o
s
ed
to
en
h
an
ce
p
r
ed
i
ctio
n
ac
cu
r
ac
y
.
T
h
is
m
o
d
el
lev
er
ag
es
m
u
ltip
le
class
if
ier
s
to
en
s
u
r
e
r
o
b
u
s
tn
ess
an
d
u
tili
ze
s
a
v
o
tin
g
m
ec
h
an
is
m
f
o
r
p
r
ed
ictio
n
a
g
g
r
eg
atio
n
,
aim
in
g
to
im
p
r
o
v
e
th
e
o
v
er
all
p
er
f
o
r
m
a
n
ce
.
T
h
e
p
r
o
ce
s
s
co
n
clu
d
es
with
an
ev
alu
atio
n
a
n
d
o
p
tim
izatio
n
s
tag
e,
wh
er
e
m
o
d
els
ar
e
f
i
n
e
-
tu
n
e
d
to
ac
h
ie
v
e
o
p
tim
al
r
esu
lts
,
d
em
o
n
s
tr
at
in
g
a
s
tr
u
ctu
r
ed
an
d
ef
f
ec
tiv
e
tech
n
iq
u
e
f
o
r
ASD
d
etec
tio
n
th
r
o
u
g
h
m
ac
h
in
e
lear
n
in
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
7
3
2
-
3
7
4
4
3736
i)
Data
u
n
d
er
s
tan
d
in
g
:
th
is
in
v
o
lv
es
ass
e
s
s
in
g
th
e
d
ataset
to
co
m
p
r
eh
e
n
d
its
co
n
ten
t,
s
tr
u
c
tu
r
e,
an
d
th
e
ty
p
es
o
f
d
ata
it
in
clu
d
es.
T
h
is
s
tu
d
y
ex
p
lo
r
e
th
e
v
ar
iab
les,
ch
ec
k
th
e
d
ata
ty
p
es,
an
d
u
n
d
er
s
tan
d
h
o
w
th
ey
co
r
r
elate
with
ea
ch
o
th
e
r
.
ii)
Data
clea
n
in
g
:
th
is
p
r
o
ce
s
s
r
ec
tifie
s
is
s
u
es
with
th
e
d
ata.
C
o
m
m
o
n
task
s
in
clu
d
e
h
an
d
lin
g
m
is
s
in
g
v
alu
es,
co
r
r
ec
tin
g
ty
p
o
s
o
r
in
ac
cu
r
ac
ies,
id
en
tify
in
g
an
d
r
e
m
o
v
in
g
d
u
p
licates,
an
d
ad
d
r
e
s
s
in
g
o
u
tlier
s
th
at
co
u
ld
s
k
ew
an
al
y
s
is
.
iii)
Data
tr
an
s
f
o
r
m
atio
n
:
h
er
e
,
t
h
e
d
ata
is
m
a
n
ip
u
lated
t
o
en
h
an
ce
t
h
e
an
aly
s
is
.
T
h
is
m
ig
h
t
in
v
o
lv
e
n
o
r
m
alizin
g
o
r
s
ca
lin
g
f
ea
t
u
r
es,
en
co
d
in
g
ca
teg
o
r
ical
v
a
r
iab
les,
cr
ea
tin
g
n
ew
f
ea
tu
r
es
f
r
o
m
e
x
is
tin
g
o
n
es,
an
d
s
u
m
m
ar
izin
g
o
r
r
est
r
u
ctu
r
in
g
d
ata
to
b
etter
f
it th
e
i
n
ten
d
ed
m
o
d
els o
r
an
aly
s
is
.
i
v
)
F
e
a
t
u
r
e
s
e
l
e
c
t
io
n
:
c
h
o
o
s
in
g
th
e
f
e
a
tu
r
e
s
t
h
a
t
a
r
e
m
o
s
t
p
e
r
ti
n
e
n
t
t
o
t
h
e
c
la
s
s
i
f
i
c
a
t
i
o
n
t
a
s
k
i
s
th
e
m
a
in
a
i
m
o
f
t
h
i
s
p
h
a
s
e
.
A
s
s
o
c
i
a
t
io
n
an
d
m
u
t
u
a
l
in
f
o
r
m
a
t
i
o
n
m
ay
b
e
u
s
e
d
in
th
i
s
w
ay
t
o
d
e
t
er
m
i
n
e
wh
i
c
h
f
e
a
t
u
r
e
s
h
a
v
e
a
r
o
b
u
s
t
a
s
s
o
c
iat
i
o
n
w
i
t
h
t
h
e
o
u
t
co
m
e
v
ar
i
ab
le
.
F
i
g
u
r
e
1
d
e
p
ic
t
s
p
r
o
p
o
s
ed
w
o
r
k
f
l
o
w
f
o
r
A
S
D
p
r
ed
i
c
t
i
o
n
.
v)
D
a
t
a
p
r
ep
r
o
c
e
s
s
i
n
g
:
i
n
th
i
s
p
h
a
s
e
,
th
e
d
a
t
a
i
s
p
r
e
-
p
r
o
ce
s
s
e
d
u
s
i
n
g
a
m
a
ch
i
n
e
l
ea
r
n
i
n
g
m
o
d
e
l
.
T
h
is
i
n
c
l
u
d
e
s
c
l
e
a
n
in
g
,
f
o
r
m
a
t
t
in
g
th
e
d
a
ta
,
f
i
l
l
in
g
i
n
th
e
m
i
s
s
i
n
g
v
a
lu
e
s
,
a
n
d
n
o
r
m
a
li
z
i
n
g
d
a
t
a
b
y
e
v
a
l
u
a
t
i
n
g
th
e
p
e
r
f
o
r
m
an
c
e
m
e
t
r
i
c
s
,
w
h
i
ch
a
r
e
e
v
a
l
u
a
t
ed
b
y
c
a
l
cu
l
a
t
i
n
g
th
e
e
f
f
e
c
t
iv
e
n
e
s
s
o
f
t
h
e
m
a
c
h
in
e
l
ea
r
n
in
g
m
o
d
e
l.
Fig
u
r
e
1
.
Pro
p
o
s
ed
wo
r
k
f
lo
w
T
h
e
m
etr
ic
u
s
ed
h
er
e
p
er
f
o
r
m
s
th
e
task
it
is
f
o
cu
s
ed
o
n
t
o
d
o
s
o
f
o
r
th
e
m
o
d
el
f
o
r
wh
ich
it
is
u
s
ed
.
W
h
er
ein
th
e
class
if
icat
io
n
m
o
d
els
ar
e
u
s
ed
f
o
r
th
e
p
u
r
p
o
s
e
o
f
ev
alu
atin
g
m
etr
ics
s
u
ch
as
ac
cu
r
ac
y
,
r
ec
all,
p
r
ec
is
io
n
,
an
d
F1
-
s
co
r
e
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
is
d
e
v
elo
p
ed
to
e
n
h
an
ce
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
b
y
estab
lis
h
in
g
an
au
tis
m
s
p
ec
tr
u
m
d
etec
tio
n
m
o
d
el
ca
p
ab
le
o
f
class
if
y
in
g
au
tis
m
.
Utilizin
g
all
th
e
m
ac
h
in
e
lear
n
in
g
class
if
ier
s
in
th
is
ex
p
er
im
en
t
is
ess
en
tial
f
o
r
en
s
u
r
in
g
th
e
b
est
r
esu
lts
f
o
r
p
r
ed
icti
n
g
au
tis
m
d
etec
tio
n
.
Alg
o
r
ith
m
1
s
h
o
ws th
e
alg
o
r
it
h
m
f
o
r
class
if
icatio
n
Alg
o
r
ith
m
1
.
Alg
o
r
ith
m
f
o
r
class
if
icatio
n
Input:
Machine_learning model, Autism_dataset
Step 1:
S
tart
Step
2
:
data <
-
load Autism_dataset
Step 3
:
If data
==
NaN
else 0
Substitute NaN or their. miss_val
Step 4
:
Pre
-
processing
If target_data = to O
Substitute O with value zero
else
Substitute D with 1
Step 5
:
z
<
-
discard {
target_data}
Step 6
:
a
<
-
target_data
Step 7
:
z1,z2,a1,a2
<
-
split (z,a) data
Step 8
:
Model
<
-
train_model using z1 and a1
Step 9
:
predict
<
-
test_model using z2 and a2
Step 10
:
u_z
<
-
data_scaling of z
Step 11
:
u_z
<
-
data_compress of u_z
Step 12:
Apply hyper_parameter tuning
for each classifier
Step 13:
Classifier
<
-
train_model using u_z
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
ea
r
ly
d
etec
tio
n
o
f
a
u
tis
m
s
p
ec
tr
u
m
d
is
o
r
d
er th
r
o
u
g
h
en
s
emb
le
-
b
a
s
ed
…
(
S
h
a
b
ee
n
a
Lyla
th
)
3737
Step 14:
Model
<
-
apply_ensemble_classifier using classifier
Step 15:
Predict
<
-
cross_validation model
Step 16:
Compute performance evaluation metrics
Step 17:
E
nd
E
x
tr
a
tr
ee
s
ar
e
th
e
ter
m
u
s
ed
to
d
escr
ib
e
th
e
DT
th
at
ar
e
p
r
o
d
u
ce
d
f
r
o
m
th
e
t
r
ain
in
g
s
et.
A
s
u
b
s
et
an
d
a
p
ar
tially
r
an
d
o
m
c
u
t
p
o
i
n
t
ar
e
r
an
d
o
m
ly
in
s
p
ec
ted
t
o
d
eter
m
in
e
a
s
p
lit
r
u
le
f
o
r
th
e
r
o
o
t
[
2
1
]
.
T
w
o
r
an
d
o
m
ly
ch
o
s
en
ch
ild
n
o
d
es
b
r
ea
k
f
r
o
m
th
e
p
ar
en
t
n
o
d
e.
U
n
til
th
e
l
ea
f
n
o
d
e
is
r
ea
ch
ed
,
ea
ch
ch
il
d
n
o
d
e
is
r
ep
ea
ted
.
T
h
e
u
ltima
te
p
r
o
jectio
n
s
ar
e
d
ec
id
ed
b
y
a
m
ajo
r
ity
v
o
te.
T
h
e
u
s
er
d
ec
id
es wh
ich
o
f
th
e
to
p
f
ea
tu
r
es to
em
p
lo
y
th
e
class
if
ier
m
o
d
el
co
n
s
id
er
e
d
as th
e
last
s
tep
,
th
is
p
r
ed
icts
th
e
d
ec
is
io
n
in
ca
s
e
o
f
r
eg
r
ess
io
n
o
r
class
if
icatio
n
p
u
r
p
o
s
e
wh
er
ein
r
eg
r
ess
io
n
in
v
o
lv
es
av
er
ag
e
p
r
ed
ictio
n
o
n
th
e
b
asis
o
f
DT
an
d
class
if
icatio
n
b
ased
o
n
tr
ee
-
b
ased
p
r
ed
ictio
n
f
o
r
m
ajo
r
ity
v
o
tin
g
.
B
ased
o
n
DT
,
th
e
L
ig
h
tGB
M
alg
o
r
ith
m
is
a
g
r
ad
ien
t
-
b
o
o
s
tin
g
m
eth
o
d
.
R
eg
r
ess
io
n
an
aly
s
is
was
u
tili
ze
d
to
class
if
y
th
e
d
ata
r
a
n
k
.
T
o
tr
ain
an
d
s
eg
r
eg
ate
th
e
d
ata
f
r
o
m
ea
ch
DT
,
two
m
eth
o
d
s
ar
e
av
ailab
le.
W
h
il
e
th
e
o
th
e
r
tech
n
iq
u
e
co
n
ce
n
tr
ates
o
n
th
e
tr
ee
'
s
leav
es,
th
e
f
ir
s
t
s
tr
ateg
y
co
n
ce
n
tr
ates
o
n
th
e
tr
ee
'
s
lev
el.
Alter
n
ativ
ely
,
to
c
o
n
tin
u
ally
s
ep
ar
ate
th
e
leav
es
an
d
r
ed
u
c
e
lo
s
s
,
a
leaf
-
wis
e
m
eth
o
d
is
u
s
ed
.
L
ig
h
tGB
M'
s
t
r
ee
s
tr
u
ctu
r
e
e
v
o
lv
es
in
a
leaf
-
wis
e
f
ash
io
n
,
with
lo
s
s
es
p
ick
ed
an
d
s
p
lit
in
ea
ch
b
r
an
ch
ac
co
r
d
in
g
to
h
o
w
m
u
c
h
th
ey
co
n
tr
ib
u
te
to
th
e
to
tal
lo
s
s
.
Fas
ter
lear
n
in
g
is
o
f
ten
a
b
en
ef
it
o
f
a
tr
ee
-
b
ased
m
o
d
el
with
a
lo
w
er
r
o
r
r
ate.
T
h
e
ev
o
lu
tio
n
o
f
th
e
L
i
g
h
tGB
M
m
o
d
el
is
m
o
s
tly
h
o
r
izo
n
tal,
p
r
ev
en
tin
g
o
v
er
lear
n
i
n
g
.
I
t
h
as b
ee
n
d
em
o
n
s
tr
ated
th
at
u
s
in
g
h
u
g
e
d
ata
s
ets p
r
o
d
u
ce
s
b
etter
r
esu
lts
.
R
id
g
e
class
if
ier
(
R
C
)
:
RC
i
s
th
e
m
eth
o
d
u
s
ed
in
m
ac
h
in
e
lear
n
in
g
t
o
ex
a
m
in
e
lin
ea
r
d
i
s
cr
im
in
an
t
m
o
d
els.
R
eg
u
lar
izatio
n
p
e
n
alize
s
th
e
co
ef
f
icien
ts
in
a
m
o
d
e
l
to
s
to
p
it
f
r
o
m
o
v
er
-
f
itti
n
g
.
Du
r
in
g
th
e
tr
ain
i
n
g
p
h
ase,
th
e
class
if
ier
f
ir
s
t
tr
ea
ts
th
e
is
s
u
e
as
a
r
e
g
r
ess
io
n
,
ch
an
g
in
g
th
e
g
o
al
v
alu
es
to
+1
an
d
-
1
.
L
DA:
a
ty
p
ical
u
s
e
o
f
th
e
d
im
en
s
io
n
al
r
ed
u
ctio
n
s
tr
ateg
y
is
in
s
u
p
e
r
v
is
ed
class
if
icatio
n
p
r
o
b
lem
s
.
Fo
r
class
if
icatio
n
,
th
e
L
DA
tech
n
iq
u
e
is
also
e
m
p
lo
y
ed
.
T
o
m
a
x
im
ize
with
i
n
-
class
v
ar
ian
ce
an
d
d
ec
r
ea
s
e
class
v
ar
iatio
n
,
th
e
L
DA
m
eth
o
d
is
u
s
ed
.
T
o
i
d
en
tify
th
e
v
ar
iab
les,
th
e
m
eth
o
d
u
s
es
a
lin
ea
r
c
o
m
b
in
atio
n
s
ea
r
ch
tech
n
i
q
u
e.
I
t
is
ass
u
r
ed
th
at
th
e
b
est d
if
f
e
r
en
ti
atin
g
f
ea
tu
r
e
f
o
r
la
b
els with
m
o
r
e
th
an
o
n
e
class
is
ch
o
s
en
[
2
2
]
.
I
n
(
1
)
to
(
4
)
ar
e
ev
al
u
ated
as
lin
ea
r
co
ef
f
icien
ts
,
th
er
e
b
y
m
a
x
im
izin
g
th
e
d
is
cr
im
in
a
n
t
f
u
n
ctio
n
.
T
h
e
eq
u
atio
n
s
h
er
e
e
r
e
p
r
esen
t
th
e
lin
ea
r
co
ef
f
icien
t
θ
m
o
d
el,
th
e
co
v
ar
ian
ce
m
atr
ix
,
an
d
t
h
e
α
s
h
o
ws
th
e
av
er
ag
e
v
ec
to
r
f
u
n
ctio
n
.
T
h
e
m
ac
h
in
e
lear
n
in
g
m
o
d
el
th
at
is
u
s
ed
in
th
e
tr
ain
in
g
p
h
ase
is
th
e
m
o
d
el
ag
g
r
eg
atio
n
-
b
ase
d
en
s
em
b
le
class
if
ier
.
Mu
ltip
le
m
o
d
els
ar
e
u
s
ed
t
o
tr
ain
th
e
class
if
ier
.
Af
ter
r
ec
eiv
in
g
th
e
o
u
tp
u
t
f
r
o
m
ea
ch
class
if
ier
,
th
e
clas
s
if
ier
f
o
r
ec
asted
th
e
o
u
tp
u
t
class
b
y
tak
in
g
in
to
ac
co
u
n
t
th
e
m
ajo
r
ity
v
o
te
with
th
e
h
ig
h
est
s
co
r
e.
E
n
s
em
b
le
m
ac
h
in
e
lear
n
in
g
m
o
d
els
f
r
eq
u
e
n
tly
u
s
e
en
s
em
b
le
ap
p
r
o
ac
h
es
to
ag
g
r
eg
a
te
p
r
ed
ictio
n
s
f
r
o
m
s
ev
er
al
in
d
ep
e
n
d
en
t
m
o
d
els.
Usi
n
g
th
e
h
ar
d
v
o
tin
g
a
p
p
r
o
a
ch
in
o
u
r
s
tu
d
y
,
th
e
class
wit
h
th
e
m
o
s
t
v
o
tes
is
d
eter
m
in
ed
b
y
ad
d
i
n
g
t
o
g
et
h
er
all
o
f
th
e
class
if
ier
s
'
p
r
ed
ictio
n
s
.
Mo
d
el
ag
g
r
eg
atio
n
-
b
ased
e
n
s
em
b
le
class
if
ier
s
ar
e
u
s
ed
to
im
p
r
o
v
e
class
if
icatio
n
r
o
b
u
s
tn
ess
an
d
ac
cu
r
ac
y
.
T
h
er
e'
s
a
ch
an
ce
th
at
s
o
m
e
d
atasets
m
ig
h
t
s
h
o
w
an
u
n
b
alan
ce
d
b
r
ea
k
d
o
wn
o
f
ca
s
es
b
y
ty
p
e.
I
t
is
cr
u
cial
to
r
em
em
b
er
th
at
tr
y
in
g
to
co
r
r
ec
tl
y
f
o
r
ec
ast
b
o
th
g
r
o
u
p
s
with
a
s
in
g
le
class
if
ier
m
ay
p
r
o
v
id
e
d
if
f
icu
lties
.
T
h
e
en
s
em
b
le
c
lass
if
ier
cr
ea
tes
a
f
o
r
ec
ast
th
at
is
m
o
r
e
th
o
r
o
u
g
h
an
d
p
r
ec
is
e
b
y
co
m
b
in
in
g
p
r
ed
ictio
n
s
f
r
o
m
s
ev
er
al
class
if
ier
s
.
Fo
u
r
b
ase
class
if
ier
s
ar
e
u
s
ed
in
th
is
s
tu
d
y
'
s
v
o
tin
g
en
s
em
b
le
m
o
d
el
.
T
h
e
ch
o
s
en
class
if
ier
f
o
r
th
is
s
y
s
tem
is
th
e
E
T
m
eta
-
class
if
ier
.
Usi
n
g
th
e
wh
o
le
tr
ain
in
g
in
p
u
t
d
ataset,
t
h
e
b
asic
m
o
d
el
was
estab
lis
h
ed
d
u
r
in
g
th
e
f
ir
s
t
tr
ain
in
g
o
f
th
e
b
asic c
lass
if
ier
s
.
T
h
e
m
eta
-
m
o
d
el
class
if
ier
r
e
ce
iv
es e
ac
h
b
ase
m
o
d
el'
s
p
r
ed
i
ctio
n
as a
n
in
p
u
t.
Usi
n
g
th
e
ad
a
p
tiv
e
e
n
s
em
b
le
class
if
ier
h
elp
s
im
p
r
o
v
e
th
e
d
ata'
s
r
esis
tan
ce
to
n
o
is
e
an
d
o
u
tlier
s
.
C
o
m
b
in
in
g
th
e
f
o
r
ec
asts
o
f
o
u
tlier
s
an
d
n
o
is
y
d
ata
m
ig
h
t
l
ess
en
th
eir
im
p
ac
t.
I
n
ad
d
itio
n
,
th
e
a
p
p
licatio
n
o
f
ad
ap
tiv
e
v
o
tin
g
en
s
em
b
le
cla
s
s
if
ier
s
in
th
e
d
etec
tio
n
an
d
ass
es
s
m
en
t
o
f
Au
tis
m
m
ay
im
p
r
o
v
e
p
r
ed
ictio
n
b
alan
ce
,
r
o
b
u
s
tn
ess
,
an
d
ac
c
u
r
ac
y
.
Alg
o
r
ith
m
2
s
h
o
ws p
s
eu
d
o
co
d
e
f
o
r
th
e
en
s
em
b
le
lear
n
i
n
g
alg
o
r
ith
m
.
=
1
1
+
1
1
+
⋯
+
(
1
)
(
)
=
1
−
2
(
2
)
(
)
=
1
−
2
(
)
=
−
1
(
1
−
2
)
(
3
)
=
1
(
1
+
2
)
(
1
1
+
2
2
)
(
4
)
Alg
o
r
ith
m
2
.
Ps
eu
d
o
co
d
e
f
o
r
th
e
en
s
em
b
le
lear
n
i
n
g
Input:
Autism training data
Output:
Ensemble classifsub
-
trained
Step 1:
Base classifiers = (ET, LGBM, RC, LDA)
Meta_classifier ET
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
7
3
2
-
3
7
4
4
3738
Step 2:
Base learner upon applying the data
-
sub
-
classifier
Step 3:
For training classifiusing cross validation
For = <
-
1 to, … do where (=10)
θ = ()
Step 4:
Learn the classifier from
End for
Step 5:
Train_set for Meta_classifier ET
Step 6:
= ()
Step 7:
Return
Step 8:
End
T
h
is
f
r
am
ewo
r
k
o
u
tlin
es
a
co
m
p
r
eh
en
s
iv
e
m
ac
h
in
e
lear
n
in
g
p
ip
elin
e
d
esig
n
e
d
f
o
r
th
e
a
n
aly
s
is
an
d
p
r
ed
ictio
n
o
f
ASD
u
s
in
g
e
n
s
em
b
le
class
if
ier
s
.
Alg
o
r
ith
m
3
s
h
o
ws
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
.
I
t
s
y
s
tem
atica
lly
p
r
o
g
r
ess
es th
r
o
u
g
h
s
ev
e
n
cr
u
c
ial
s
tep
s
.
Fig
u
r
e
2
p
r
esen
ts
th
e
f
lo
wch
ar
t o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
.
Alg
o
r
ith
m
3
.
T
h
e
p
r
o
p
o
s
ed
al
g
o
r
ith
m
Step 1:
Preliminary analysis
for dataset in [ ASD_Dataset]:
if dataset.is_available ():
analyze_dataset_characteristics(dataset)
Step 2:
Data acquisition
ASD_Data = acquire data ('ASD _Dataset')
ASD_Data = acquire data('ASD_Dataset')
combined data = combine datasets(ASD_children_Data, ASD_ Adolescents_Data)
Step 3:
Data pre
-
processing
for data in combined data:
if data. needs conversion ():
convert_categorical_to_numerical(data)
if data.is_imbalanced ():
apply_(data)
Step 4:
Feature Selection
selected_features = []
for feature in combined data. features:
if is_relevant_feature(feature):
selected_features. append(feature)
apply_feature_selection_methods
(feature, methods=[,
Extra_Trees_Classifier])
Step 5
:
Ensemble_classifier model implementation
trained_models = {}
for model in []:
trained_models[model.name] = train_model(model, selected_features)
Step 6
:
Ensemble learning
ensemble_models = []
for model in []:
if model in trained_models:
ensemble_models. append(trained_models[model])
ensemble = create ensemble (ensemble_models, method='soft_voting')
Step 7
:
Evaluation and optimization
for model in [ensemble] + list (trained_models. Values ()):
evaluate_model_performance(model)
if model. needs optimization ():
apply_hyperparameter_tuning(model)
Pre
lim
in
ar
y
an
aly
s
is
:
t
h
e
p
ip
e
lin
e
s
tar
ts
b
y
ass
ess
in
g
th
e
av
ailab
ilit
y
an
d
ch
ar
ac
te
r
is
tics
o
f
an
ASD
d
ataset.
T
h
is
f
o
u
n
d
atio
n
al
s
tep
en
s
u
r
es
a
d
ee
p
u
n
d
er
s
tan
d
in
g
o
f
th
e
d
ataset's
s
tr
u
ctu
r
e,
q
u
ality
,
an
d
r
elev
a
n
ce
to
ASD
b
ef
o
r
e
p
r
o
ce
ed
i
n
g
f
u
r
th
er
.
Data
ac
q
u
is
itio
n
:
it
th
e
n
m
o
v
es
o
n
to
g
ath
e
r
ASD
-
r
elate
d
d
ata,
p
o
te
n
tially
f
r
o
m
v
ar
io
u
s
s
o
u
r
ce
s
,
to
f
o
r
m
a
r
ich
d
ataset.
T
h
is
s
tep
m
ig
h
t
in
clu
d
e
co
m
b
in
in
g
d
ata
f
r
o
m
s
p
ec
if
ic
s
u
b
s
ets,
s
u
ch
as
ch
ild
r
en
an
d
ad
o
lesce
n
ts
with
ASD,
to
en
s
u
r
e
a
d
iv
er
s
e
an
d
co
m
p
r
eh
en
s
iv
e
d
ataset
f
o
r
an
aly
s
is
.
Data
p
r
e
-
p
r
o
ce
s
s
in
g
:
th
e
g
ath
er
e
d
d
ata
u
n
d
er
g
o
es
cr
itical
p
r
e
-
p
r
o
ce
s
s
in
g
task
s
,
in
clu
d
in
g
th
e
co
n
v
er
s
io
n
o
f
ca
teg
o
r
ical
d
ata
in
t
o
a
n
u
m
er
ical
f
o
r
m
at
f
o
r
m
ac
h
in
e
lear
n
in
g
co
m
p
atib
ilit
y
an
d
ad
d
r
ess
in
g
an
y
im
b
alan
ce
with
in
th
e
d
ataset
to
p
r
ev
e
n
t
b
iased
m
o
d
el
p
r
ed
ictio
n
s
.
Featu
r
e
s
elec
tio
n
:
id
en
tif
y
in
g
a
n
d
s
elec
tin
g
th
e
m
o
s
t
in
f
o
r
m
ativ
e
f
ea
t
u
r
es
is
k
ey
to
im
p
r
o
v
in
g
m
o
d
el
p
er
f
o
r
m
an
c
e.
T
h
is
s
tep
em
p
lo
y
s
b
o
th
m
a
n
u
al
in
s
p
ec
tio
n
an
d
au
to
m
ated
m
eth
o
d
s
,
s
u
ch
as th
e
E
T
class
if
ier
,
to
d
eter
m
in
e
t
h
e
m
o
s
t r
elev
an
t
f
ea
tu
r
es f
o
r
ASD
p
r
ed
ictio
n
.
E
n
s
em
b
le
class
if
ier
m
o
d
el
im
p
lem
en
tatio
n
:
v
ar
i
o
u
s
m
ac
h
in
e
lear
n
in
g
m
o
d
els
ar
e
tr
ai
n
ed
o
n
th
e
s
elec
ted
f
ea
tu
r
es.
T
h
is
s
tep
allo
ws
f
o
r
th
e
f
lex
ib
ilit
y
to
ch
o
o
s
e
th
e
m
o
s
t
ap
p
r
o
p
r
iate
m
o
d
els
b
ased
o
n
t
h
e
d
ata'
s
ch
ar
ac
ter
is
tics
an
d
th
e
p
r
elim
in
ar
y
a
n
aly
s
is
's
in
s
ig
h
ts
.
E
n
s
em
b
le
lear
n
in
g
:
e
n
h
an
cin
g
th
e
p
r
e
d
ictiv
e
p
er
f
o
r
m
an
ce
b
y
co
m
b
in
in
g
th
e
s
tr
en
g
th
s
o
f
in
d
iv
id
u
al
m
o
d
els
th
r
o
u
g
h
an
en
s
em
b
le
ap
p
r
o
ac
h
,
s
p
ec
if
ically
u
s
in
g
s
o
f
t
v
o
tin
g
.
T
h
is
m
eth
o
d
ag
g
r
e
g
ates
th
e
p
r
ed
ictio
n
s
o
f
m
u
ltip
le
m
o
d
els
to
m
ak
e
a
f
in
al
p
r
ed
ictio
n
,
lev
er
ag
in
g
th
eir
c
o
llectiv
e
in
tellig
en
ce
f
o
r
im
p
r
o
v
ed
ac
cu
r
ac
y
.
E
v
alu
atio
n
an
d
o
p
tim
izatio
n
:
th
e
last
s
tep
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
ea
r
ly
d
etec
tio
n
o
f
a
u
tis
m
s
p
ec
tr
u
m
d
is
o
r
d
er th
r
o
u
g
h
en
s
emb
le
-
b
a
s
ed
…
(
S
h
a
b
ee
n
a
Lyla
th
)
3739
in
v
o
lv
e
a
th
o
r
o
u
g
h
ev
alu
atio
n
o
f
th
e
p
er
f
o
r
m
an
ce
o
f
b
o
th
t
h
e
en
s
em
b
le
an
d
in
d
iv
id
u
al
m
o
d
els,
f
o
llo
wed
b
y
h
y
p
er
p
ar
am
eter
tu
n
i
n
g
to
o
p
t
im
ize
th
eir
ef
f
ec
tiv
en
ess
.
T
h
is
iter
ativ
e
p
r
o
ce
s
s
en
s
u
r
es
th
at
th
e
m
o
d
els
ar
e
f
in
ely
tu
n
ed
f
o
r
th
e
b
est
p
o
s
s
ib
le
p
r
e
d
ictiv
e
p
er
f
o
r
m
a
n
ce
o
n
ASD.
T
h
r
o
u
g
h
o
u
t
th
is
p
r
o
ce
s
s
,
th
e
em
p
h
asis
is
o
n
d
ata
p
r
ep
ar
atio
n
,
s
tr
ateg
ic
m
o
d
el
s
elec
tio
n
,
an
d
c
o
n
tin
u
o
u
s
r
ef
in
em
e
n
t
o
f
t
h
e
m
o
d
els.
T
h
is
ap
p
r
o
ac
h
aim
s
to
lev
er
ag
e
th
e
s
tr
en
g
t
h
s
o
f
e
n
s
em
b
le
lear
n
in
g
to
ac
c
u
r
atel
y
p
r
e
d
ict
ASD,
d
em
o
n
s
tr
atin
g
a
s
o
p
h
is
ticated
an
d
th
o
u
g
h
t
f
u
l a
p
p
licatio
n
o
f
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es to
a
c
o
m
p
lex
a
n
d
s
en
s
itiv
e
m
ed
ical
co
n
d
itio
n
.
I
m
p
lem
en
tatio
n
s
ettin
g
s
:
t
h
e
p
r
o
p
o
s
ed
en
s
em
b
le
was
im
p
lem
en
ted
in
Py
th
o
n
u
s
in
g
th
e
S
cik
it
-
lear
n
an
d
L
ig
h
tGB
M
lib
r
ar
ies.
T
h
e
en
s
em
b
le
co
m
b
in
es
f
o
u
r
b
ase
lear
n
er
s
—
E
T
,
L
ig
h
tGB
M
,
R
C
,
an
d
L
DA
—
ag
g
r
eg
ated
t
h
r
o
u
g
h
a
h
a
r
d
(
m
ajo
r
ity
)
v
o
tin
g
s
ch
em
e,
with
th
e
E
T
class
if
ier
also
s
er
v
in
g
as th
e
m
eta
-
class
if
ier
.
C
ateg
o
r
ical
attr
ib
u
tes
wer
e
la
b
el
-
en
co
d
e
d
,
m
is
s
in
g
v
alu
es
wer
e
im
p
u
ted
,
an
d
f
ea
t
u
r
es
w
er
e
n
o
r
m
alize
d
p
r
io
r
to
tr
ain
in
g
;
th
e
m
o
s
t
in
f
o
r
m
a
tiv
e
f
ea
tu
r
es
wer
e
s
elec
ted
u
s
in
g
th
e
E
T
f
ea
tu
r
e
-
im
p
o
r
tan
ce
r
an
k
in
g
.
M
o
d
el
p
er
f
o
r
m
an
ce
was
esti
m
ated
u
s
in
g
1
0
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
o
n
th
e
p
u
b
licly
a
v
ailab
le
ch
il
d
r
en
an
d
ad
u
lt
ASD
s
cr
ee
n
in
g
d
atasets
,
an
d
r
ep
o
r
ted
in
ter
m
s
o
f
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e.
Def
au
lt
lib
r
ar
y
h
y
p
er
p
ar
am
eter
s
wer
e
u
s
ed
f
o
r
th
e
b
ase
lear
n
e
r
s
u
n
less
o
th
er
wis
e
s
tated
,
en
s
u
r
in
g
th
e
p
ip
elin
e
ca
n
b
e
r
ep
r
o
d
u
ce
d
b
y
o
t
h
er
r
esear
c
h
e
r
s
.
Fig
u
r
e
2
.
Flo
wch
ar
t
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
it
h
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
7
3
2
-
3
7
4
4
3740
4.
P
E
RF
O
RM
A
NCE
E
VA
L
U
AT
I
O
N
T
h
e
p
er
f
o
r
m
an
ce
ev
alu
atio
n
is
ca
r
r
ied
o
u
t
u
s
in
g
th
e
two
d
atasets
,
th
e
ch
ild
r
en
d
atas
et
an
d
th
e
ad
o
lescen
t
d
ataset,
wh
er
ein
t
h
e
ex
is
tin
g
s
tate
-
of
-
th
e
-
ar
t
te
ch
n
iq
u
es
ar
e
c
o
m
p
a
r
ed
with
t
h
e
p
r
o
p
o
s
ed
m
o
d
el,
an
d
th
e
r
esu
lts
ar
e
g
iv
en
in
th
e
f
o
r
m
o
f
g
r
ap
h
s
,
as
s
h
o
wn
in
T
ab
le
1
an
d
Fig
u
r
e
3
.
Fig
u
r
e
3
(
a)
p
r
esen
ts
ac
cu
r
ac
y
,
Fig
u
r
e
3
(
b
)
p
r
esen
t
s
p
r
ec
is
io
n
,
Fig
u
r
e
3
(
c)
p
r
ese
n
ts
r
ec
all,
an
d
Fig
u
r
e
3
(
d
)
p
r
esen
ts
F1
-
s
co
r
e.
I
n
th
is
s
tu
d
y
,
p
e
o
p
le
with
ASD
wh
o
ar
e
d
iv
id
ed
in
to
two
ag
e
g
r
o
u
p
s
,
ch
ild
r
en
a
n
d
a
d
u
lts
,
ar
e
d
r
awn
f
r
o
m
two
d
if
f
er
en
t
d
atasets
.
A
1
0
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
tech
n
i
q
u
e
i
s
u
s
ed
to
b
u
ild
p
r
ed
ictio
n
m
o
d
els
u
s
in
g
th
e
d
atasets
[
2
3
]
,
[
2
4
]
.
I
n
th
e
1
0
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
tr
ain
in
g
p
h
ase,
th
e
d
atasets
ar
e
s
p
lit
i
n
to
ten
eq
u
al
-
s
ized
g
r
o
u
p
s
,
o
r
f
o
ld
s
,
at
r
an
d
o
m
.
On
e
-
f
o
ld
is
r
eser
v
ed
f
o
r
te
s
tin
g
in
th
e
m
o
d
el
-
b
u
ild
in
g
p
r
o
ce
s
s
,
wh
ile
th
e
r
em
ain
in
g
n
in
e
f
o
ld
s
ar
e
u
s
ed
f
o
r
tr
ain
in
g
.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
3
.
Per
f
o
r
m
an
c
e
co
m
p
ar
is
o
n
o
n
ch
ild
r
en
d
ataset
: (
a)
ac
cu
r
ac
y
,
(
b
)
p
r
ec
is
io
n
,
(
c
)
r
ec
al
l,
an
d
(
d
)
F1
-
s
co
r
e
T
h
e
tech
n
iq
u
e
is
r
ep
ea
ted
n
u
m
er
o
u
s
tim
es,
p
ar
ticu
lar
ly
1
0
ti
m
es,
an
d
th
e
av
e
r
ag
e
o
f
th
o
s
e
r
ep
etitio
n
s
is
u
s
ed
to
d
eter
m
in
e
th
e
f
in
d
in
g
s
.
I
n
th
is
in
s
tan
ce
,
a
1
0
-
f
o
l
d
cr
o
s
s
-
v
alid
atio
n
tech
n
iq
u
e
is
u
s
ed
in
th
e
p
r
o
ce
s
s
to
p
r
o
v
id
e
a
m
o
r
e
wid
ely
ap
p
licab
le
m
o
d
el.
T
h
is
is
a
v
er
y
u
s
ef
u
l
r
eso
u
r
ce
wh
en
s
p
ar
s
e
d
ata
is
p
r
esen
t,
a
n
d
o
v
er
f
itti
n
g
n
ee
d
s
to
b
e
m
in
i
m
ized
.
Pre
v
en
ti
n
g
o
v
er
f
itti
n
g
is
ac
co
m
p
lis
h
ed
b
y
l
o
wer
in
g
th
e
v
ar
ian
ce
w
h
ile
cr
ea
tin
g
th
e
m
o
d
el.
T
h
e
lack
o
f
ad
e
q
u
ate
s
am
p
les
in
th
e
d
a
tasets
m
ay
b
e
th
e
ca
u
s
e
o
f
t
h
is
v
ar
ian
ce
d
ec
r
ea
s
e.
W
h
en
h
o
ld
o
u
t
v
alid
atio
n
is
u
s
ed
d
u
r
in
g
m
o
d
el
d
ev
elo
p
m
e
n
t
with
a
f
ix
ed
test
s
et,
o
v
er
f
i
ttin
g
m
ig
h
t
h
a
p
p
en
.
Ov
er
f
itti
n
g
is
a
p
h
en
o
m
en
o
n
t
h
at
ca
n
ca
u
s
e
v
ar
ian
ce
t
o
r
is
e,
m
ak
in
g
it
m
o
r
e
d
if
f
icu
lt
f
o
r
th
e
p
r
ed
ictio
n
m
o
d
el
to
g
en
er
alize
to
test
d
ata
th
a
t
h
asn
'
t
b
ee
n
s
ee
n
b
ef
o
r
e.
Sev
er
al
s
tatis
tica
l
as
s
es
s
m
en
t
m
etr
ics
ar
e
u
s
ed
to
v
alid
ate
th
e
ex
p
er
im
en
tal
o
u
tco
m
es.
Acc
u
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
an
d
r
ec
eiv
er
o
p
er
atin
g
ch
ar
ac
ter
is
tics
(
R
O
C
)
cu
r
v
e
ar
e
am
o
n
g
m
etr
ics
[
2
5
]
–
[
2
7
]
.
T
h
e
r
e
s
u
l
t
s
ar
e
e
v
a
lu
a
t
ed
o
n
t
w
o
d
a
t
a
s
e
t
s
,
t
h
e
ch
i
l
d
r
e
n
a
n
d
ad
u
l
t
d
a
t
a
s
e
t
f
o
r
th
e
p
er
f
o
r
m
an
c
e
m
e
t
r
i
c
s
a
c
c
u
r
a
cy
,
F
1
-
s
co
r
e,
p
r
e
c
i
s
i
o
n
,
a
n
d
r
e
c
a
l
l
.
T
h
e
r
e
s
u
l
t
s
a
r
e
s
h
o
wn
in
th
e
f
o
r
m
o
f
a
g
r
a
p
h
a
n
d
t
ab
l
e
s
.
T
h
e
d
a
t
a
s
e
t
o
v
e
r
v
i
e
w
f
o
r
AS
D
p
r
e
d
i
c
ti
o
n
in
c
h
i
l
d
r
e
n
s
h
o
w
c
a
s
e
s
a
s
e
r
i
e
s
o
f
s
t
u
d
i
e
s
,
e
a
c
h
co
n
tr
i
b
u
t
i
n
g
t
o
t
h
e
ev
o
lv
i
n
g
l
an
d
s
c
ap
e
o
f
d
i
a
g
n
o
s
t
i
c
a
c
cu
r
ac
y
th
r
o
u
g
h
m
ac
h
in
e
l
e
ar
n
in
g
m
o
d
e
l
s
.
N
o
t
ab
l
y
,
th
e
p
r
o
g
r
e
s
s
i
o
n
i
n
m
o
d
e
l
p
e
r
f
o
r
m
a
n
c
e
i
s
ev
i
d
en
t
,
cu
l
m
in
a
t
in
g
i
n
th
e
p
r
o
p
o
s
ed
m
o
d
e
l,
w
h
i
c
h
a
ch
i
ev
e
s
t
h
e
h
i
g
h
e
s
t
s
c
o
r
e
s
ac
r
o
s
s
a
l
l
m
e
t
r
i
c
s
:
a
n
a
c
cu
r
a
cy
o
f
9
8
.
9
6
%
,
an
F
1
-
s
c
o
r
e
o
f
9
8
.
6
7
%
,
p
r
e
c
i
s
io
n
a
t
9
7
.
8
7
%
,
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
ci
n
g
ea
r
ly
d
etec
tio
n
o
f
a
u
tis
m
s
p
ec
tr
u
m
d
is
o
r
d
er th
r
o
u
g
h
en
s
emb
le
-
b
a
s
ed
…
(
S
h
a
b
ee
n
a
Lyla
th
)
3741
r
e
c
a
l
l
a
t
9
8
.
3
2
%
.
T
h
i
s
i
n
d
i
c
at
e
s
a
s
o
p
h
i
s
t
i
c
a
t
ed
m
o
d
e
l
c
ap
a
b
l
e
o
f
a
c
cu
r
a
t
e
ly
id
e
n
t
if
y
in
g
A
S
D
ca
s
e
s
w
i
t
h
m
i
n
i
m
a
l
f
a
l
s
e
p
o
s
i
t
i
v
e
s
(
h
ig
h
p
r
e
c
i
s
i
o
n
)
an
d
f
a
l
s
e
n
e
g
a
ti
v
e
s
(
h
ig
h
r
e
c
a
l
l)
,
s
h
o
w
c
a
s
in
g
th
e
p
o
t
en
t
i
a
l
o
f
a
d
v
an
c
ed
a
lg
o
r
i
t
h
m
s
i
n
en
h
an
c
i
n
g
A
S
D
d
i
a
g
n
o
s
t
i
c
s
.
E
a
r
li
e
r
co
n
tr
i
b
u
t
i
o
n
s
,
l
i
k
e
B
a
l
a
e
t
a
l
.
[
1
]
w
i
th
an
a
c
c
u
r
a
c
y
o
f
9
9
.
6
1
%
an
d
a
n
F
1
-
s
c
o
r
e
o
f
9
9
.
6
,
a
n
d
H
ar
o
o
n
an
d
P
a
d
m
a
[
2
8
]
w
i
t
h
b
a
l
an
c
e
d
p
r
e
c
i
s
i
o
n
a
n
d
r
e
c
a
l
l
,
s
e
t
s
ig
n
i
f
ic
a
n
t
b
en
ch
m
a
r
k
s
.
H
o
we
v
er
,
t
h
e
p
r
o
p
o
s
e
d
m
o
d
e
l
s
u
r
p
a
s
s
e
s
t
h
e
s
e
t
h
r
o
u
g
h
i
n
c
r
em
e
n
t
a
l
i
m
p
r
o
v
em
e
n
t
s
i
n
b
a
l
an
c
i
n
g
t
h
e
p
r
e
c
i
s
i
o
n
-
r
ec
a
l
l
t
r
ad
e
-
o
f
f
,
h
i
g
h
l
i
g
h
t
i
n
g
th
e
c
o
n
t
i
n
u
o
u
s
r
e
f
in
em
e
n
t
an
d
p
o
t
en
t
i
a
l
o
f
m
a
c
h
in
e
le
a
r
n
i
n
g
i
n
c
o
n
t
r
ib
u
t
in
g
to
ef
f
e
c
t
iv
e,
e
a
r
ly
A
S
D
d
e
t
e
c
t
io
n
[
2
9
]
–
[
3
2
]
.
T
ab
l
e
1
s
h
o
w
s
t
h
e
ch
i
l
d
r
e
n
d
a
t
a
s
e
t.
T
ab
le
1
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
o
n
th
e
ch
ild
r
e
n
d
ataset
R
e
f
e
r
e
n
c
e
A
c
c
u
r
a
c
y
F1
-
sc
o
r
e
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
O
mar
e
t
a
l
.
[
2
5
]
9
2
.
2
6
Th
a
b
t
a
h
[
2
6
]
9
7
.
8
98
Ta
l
a
b
a
n
i
a
n
d
A
v
c
i
[
2
7
]
9
2
.
2
6
8
8
.
0
9
9
6
.
5
2
B
a
l
a
e
t
a
l
.
[
1
]
9
9
.
6
1
9
9
.
6
H
a
r
o
o
n
a
n
d
P
a
d
ma
[
2
8
]
9
5
.
5
96
97
98
A
b
i
t
h
a
e
t
a
l
.
[
2
9
]
9
4
.
1
K
a
m
ma
e
t
a
l
.
[
3
0
]
9
5
.
8
2
G
u
p
t
a
e
t
a
l
.
[
3
1
]
9
4
.
7
1
9
2
.
5
9
9
7
.
0
9
Ex
i
s
t
i
n
g
m
o
d
e
l
[
3
2
]
9
7
.
9
5
9
7
.
0
2
9
6
.
1
6
9
7
.
7
2
P
r
o
p
o
se
d
m
o
d
e
l
9
8
.
9
6
9
8
.
6
7
9
7
.
8
7
9
8
.
3
2
T
h
is
d
ataset
o
u
tlin
es
th
e
a
d
v
an
ce
m
en
ts
in
m
ac
h
in
e
lear
n
in
g
m
o
d
els
f
o
r
a
s
p
ec
i
f
ic
ap
p
licatio
n
,
ev
id
en
ce
d
b
y
p
r
o
g
r
ess
iv
ely
im
p
r
o
v
i
n
g
m
etr
ics
ac
r
o
s
s
v
ar
io
u
s
s
tu
d
ies.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
s
ets
a
n
ew
b
en
ch
m
ar
k
with
n
ea
r
ly
p
er
f
ec
t
s
co
r
es:
an
ac
cu
r
ac
y
o
f
9
9
.
8
7
%,
an
F1
-
s
co
r
e
o
f
9
9
.
9
8
%,
p
r
ec
is
io
n
at
9
9
.
8
7
%,
an
d
a
r
ec
all
o
f
1
0
0
%.
Su
ch
h
ig
h
p
e
r
f
o
r
m
an
ce
in
d
icate
s
an
e
x
ce
p
tio
n
al
ca
p
ab
ilit
y
in
i
d
en
ti
f
y
in
g
t
r
u
e
p
o
s
itiv
es
with
o
u
t
in
cr
ea
s
in
g
f
alse
p
o
s
itiv
es
o
r
n
eg
ativ
es,
s
h
o
wca
s
in
g
a
s
ig
n
if
ican
t
leap
in
p
r
ed
ictiv
e
ac
cu
r
ac
y
.
T
h
e
jo
u
r
n
ey
to
war
d
s
th
is
p
i
n
n
ac
le
o
f
m
o
d
el
p
e
r
f
o
r
m
an
ce
i
n
clu
d
es
n
o
tewo
r
th
y
m
iles
to
n
es,
s
u
ch
as
T
h
ab
tah
[
2
6
]
ac
h
iev
in
g
an
alm
o
s
t
p
er
f
ec
t
r
ec
all
o
f
9
9
.
9
%
an
d
B
ala
et
a
l
.
[
1
]
p
r
esen
tin
g
an
o
u
ts
tan
d
in
g
b
alan
ce
ac
r
o
s
s
all
m
etr
ics
with
an
ac
cu
r
ac
y
an
d
F1
-
s
co
r
e
clo
s
e
to
1
0
0
%.
T
h
e
g
r
ad
u
al
in
cr
ea
s
e
in
m
o
d
el
m
etr
ics
f
r
o
m
ea
r
lier
s
tu
d
ies
lik
e
[
2
5
]
,
[
3
3
]
to
th
e
p
r
o
p
o
s
ed
m
o
d
el
h
ig
h
lig
h
ts
th
e
im
p
ac
tf
u
l
p
r
o
g
r
ess
in
th
e
f
ield
,
d
r
iv
en
b
y
co
n
tin
u
o
u
s
m
eth
o
d
o
lo
g
ical
r
e
f
in
em
en
ts
an
d
tec
h
n
o
lo
g
ical
ad
v
an
ce
m
e
n
ts
.
T
h
is
ev
o
lu
tio
n
r
ef
lects
a
f
o
cu
s
ed
ef
f
o
r
t
to
war
d
s
ac
h
iev
in
g
n
ea
r
-
p
er
f
ec
t
m
o
d
el
p
er
f
o
r
m
an
ce
,
u
n
d
er
s
co
r
in
g
th
e
p
o
ten
tial
o
f
cu
r
r
en
t
m
ac
h
i
n
e
lear
n
in
g
ap
p
r
o
ac
h
es
i
n
r
ea
c
h
i
n
g
u
n
p
r
ec
ed
en
ted
lev
els
o
f
p
r
ec
is
io
n
in
th
eir
a
p
p
licatio
n
s
.
T
h
e
co
m
p
ar
ativ
e
r
esu
lts
f
o
r
th
e
ad
u
lt
d
ataset
ar
e
p
r
esen
ted
in
T
ab
le
2
an
d
Fi
g
u
r
e
4
.
Fig
u
r
e
4
(
a
)
p
r
esen
ts
a
cc
u
r
ac
y
,
Fig
u
r
e
4
(
b
)
p
r
esen
ts
p
r
ec
is
io
n
,
Fig
u
r
e
4
(
c)
p
r
esen
ts
r
ec
all,
an
d
Fig
u
r
e
4
(
d
)
p
r
esen
ts
F1
-
s
co
r
e.
I
n
th
e
co
m
p
ar
is
o
n
an
aly
s
is
o
f
p
er
f
o
r
m
a
n
ce
m
etr
ics
b
etwe
en
th
e
ex
is
tin
g
s
y
s
tem
an
d
th
e
p
r
o
p
o
s
e
d
s
y
s
tem
f
o
r
b
o
t
h
ch
ild
r
e
n
an
d
ad
u
lt
d
atasets
,
s
ig
n
if
ican
t
im
p
r
o
v
em
e
n
ts
ar
e
o
b
s
er
v
ed
ac
r
o
s
s
all
k
ey
m
etr
ics,
in
d
icatin
g
t
h
e
ef
f
icac
y
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
.
Fo
r
th
e
ch
ild
r
en
d
ataset,
th
e
p
r
o
p
o
s
e
d
s
y
s
tem
s
h
o
ws
a
n
en
h
an
ce
m
e
n
t
in
ac
cu
r
ac
y
b
y
1
.
0
3
%,
p
r
ec
is
io
n
b
y
1
.
7
1
%,
a
n
d
r
ec
all
b
y
0
.
6
1
%,
with
th
e
F1
-
s
co
r
e
s
ee
in
g
a
n
o
tab
le
r
is
e
o
f
1
.
7
%.
T
h
is
in
d
icate
s
a
m
o
r
e
ac
cu
r
ate
an
d
r
el
iab
le
d
etec
tio
n
o
f
ASD
in
y
o
u
n
g
er
p
o
p
u
latio
n
s
.
Similar
ly
,
th
e
ad
u
lt
d
ataset
co
m
p
ar
is
o
n
r
e
v
ea
ls
s
u
b
s
tan
tial
im
p
r
o
v
e
m
en
ts
,
p
ar
ticu
lar
ly
i
n
p
r
ec
is
io
n
,
wh
ich
ju
m
p
s
b
y
1
.
7
1
%,
a
n
d
th
e
F1
-
s
co
r
e,
wh
ic
h
in
c
r
ea
s
es
s
lig
h
tly
b
y
0
.
7
7
%,
wh
ile
m
ain
tain
i
n
g
a
p
er
f
ec
t
r
ec
all
r
ate.
Acc
u
r
ac
y
also
s
ee
s
a
c
o
n
s
id
e
r
ab
le
b
o
o
s
t
o
f
0
.
8
4
%.
T
h
ese
im
p
r
o
v
em
e
n
ts
u
n
d
er
s
co
r
e
t
h
e
p
r
o
p
o
s
ed
s
y
s
tem
'
s
ca
p
ab
ilit
y
to
p
r
o
v
id
e
m
o
r
e
p
r
ec
is
e
an
d
d
ep
en
d
ab
le
d
iag
n
o
s
es
ac
r
o
s
s
d
if
f
er
e
n
t
ag
e
g
r
o
u
p
s
,
af
f
ir
m
in
g
its
p
o
ten
tial to
s
ig
n
if
ican
tly
im
p
a
ct
ASD
d
etec
tio
n
an
d
in
ter
v
en
tio
n
s
tr
ateg
ies.
T
ab
le
2
.
Per
f
o
r
m
an
ce
m
etr
ics
o
n
th
e
a
d
u
lt d
ataset
R
e
f
e
r
e
n
c
e
A
c
c
u
r
a
c
y
F1
-
sc
o
r
e
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
O
mar
e
t
a
l
.
[
2
5
]
9
7
.
1
Th
a
b
t
a
h
[
2
6
]
9
9
.
8
5
9
9
.
9
S
h
u
v
o
e
t
a
l
.
[
3
3
]
9
5
.
7
1
8
5
.
7
1
Ta
l
a
b
a
n
i
a
n
d
A
v
c
i
[
2
7
]
9
6
.
9
1
9
0
.
0
7
9
6
.
8
7
B
a
l
a
e
t
a
l
.
[
1
]
9
9
.
8
2
9
9
.
9
9
9
.
9
A
b
i
t
h
a
e
t
a
l
.
[
2
9
]
98
K
a
m
ma
e
t
a
l
.
[
3
0
]
9
5
.
8
2
G
u
p
t
a
e
t
a
l
.
[
3
1
]
9
4
.
2
6
9
7
.
4
6
9
1
.
2
7
Ex
i
s
t
i
n
g
m
o
d
e
l
[
3
2
]
9
9
.
0
3
9
9
.
1
1
9
8
.
1
6
1
0
0
P
r
o
p
o
se
d
m
o
d
e
l
9
9
.
8
7
9
9
.
9
8
9
9
.
8
7
1
0
0
Evaluation Warning : The document was created with Spire.PDF for Python.