I
A
E
S
I
n
t
e
r
n
at
io
n
al
Jou
r
n
al
of
A
r
t
if
ic
ia
l
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
V
ol
.
14
, N
o.
5
,
O
c
to
be
r
2025
, pp.
3681
~
3692
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
14
.i
5
.pp
3681
-
3692
3681
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
O
p
t
i
m
i
z
i
n
g d
i
a
b
e
t
e
s p
r
e
d
i
c
t
i
on
:
u
n
ve
i
l
i
n
g p
at
i
e
n
t
s
u
b
gr
ou
p
s
t
h
r
ou
gh
c
l
u
st
e
r
i
n
g
R
it
a G
an
gu
ly
1
,
D
h
ar
m
p
al
S
in
gh
2
,
R
a
j
e
s
h
B
os
e
2
1
D
e
pa
r
t
m
e
nt
of
C
om
put
e
r
A
ppl
i
c
a
t
i
ons
, D
r
.
BC
R
oy A
c
a
de
m
y of
P
r
of
e
s
s
i
ona
l
C
our
s
e
s
, D
ur
ga
pur
,
I
ndi
a
2
D
e
pa
r
t
m
e
nt
of
C
om
put
e
r
S
c
i
e
nc
e
,
F
a
c
ul
t
y of
E
ngi
ne
e
r
i
ng a
nd T
e
c
hnol
ogy
,
J
I
S
U
ni
ve
r
s
i
t
y, K
ol
ka
t
a
, I
ndi
a
A
r
t
ic
le
I
n
f
o
A
B
S
T
R
A
C
T
A
r
ti
c
le
h
is
to
r
y
:
R
e
c
e
iv
e
d
J
ul
17, 2024
R
e
vi
s
e
d
J
un 24, 2025
A
c
c
e
pt
e
d
J
ul
13, 2025
Diabetes
is
a
significant
global
health
concern,
leading
to
numerous
deaths
annually
and
affecting
many
indivi
duals
who
remain
undiagno
sed.
As
its
prevalence
rises,
the
importance
of
early
detection
becomes
incre
asingly
vital.
The
rising
diabetes
epidemic
demands
data
-
driven
strategies
to
catch
health
problems
sooner
and
identify
them
clearly.
This
study
utiliz
es
the
Pima
Indians
diabetes
dataset
(PIDD)
to
compare
three
powerful
c
lu
stering
schemes
such
as
k
-
means,
fuzzy
C
-
means,
and
hierarchical.
Uncon
troll
ed
diabetes,
arising
from
the
body'
s
struggle
to
manage
blood
sugar
due
to
insulin
deficienc
y,
can
lead
to
devastating
complications.
Early
de
tection
and
intervention
are
the
cornerstones
of
effective
management
and
im
proved
patient
outcomes.
This
study
breaks
new
ground
by
meticulously
eval
uating
the
performance
of
each
clustering
algorithm
using
advanced
metri
cs
like
silhouette
score
and
adjusted
Rand
index.
The
goal
is
to
identify
the
method
that
generates
the
most
accurate
and
well
-
defined
clusters
for
di
abetes
-
related
attributes
.
This,
in
turn,
has
the
potential
to
revolution
ize
di
abetes
diagnosis,
enabling
earlier
interventions
and
ultimately
leading
to
better
disease
management
and
patient
care.
By
providing
a
compreh
ensive
compariso
n
of
these
clusteri
ng
techniqu
es,
this
research
offers
a
sign
ificant
contribu
tion t
o the fi
ght agai
nst di
abetes.
K
e
y
w
o
r
d
s
:
C
lu
s
te
r
in
g m
e
th
od
D
ia
be
te
s
F
uz
z
y C
-
m
e
a
ns
H
ie
r
a
r
c
hi
c
a
l
c
lu
s
te
r
in
g
K
-
m
e
a
ns
P
im
a
di
a
be
te
s
da
ta
s
e
t
S
il
houe
tt
e
s
c
or
e
This is an
open
acce
ss artic
le unde
r the
CC BY
-
SA
license.
C
or
r
e
s
pon
di
n
g A
u
th
or
:
R
it
a
G
a
ngul
y
D
e
pa
r
tm
e
nt
of
C
om
put
e
r
A
ppl
ic
a
ti
ons
, D
r
. B
C
R
oy
A
c
a
de
m
y o
f
P
r
of
e
s
s
io
na
l
C
our
s
e
s
F
ul
jh
or
e
, D
ur
ga
pur
(
P
a
c
c
hi
m
B
ur
dw
a
n)
–
713206, I
ndi
a
E
m
a
il
:
ga
ngul
y.r
it
a
@
gm
a
il
.c
om
1.
I
N
T
R
O
D
U
C
T
I
O
N
H
e
a
lt
hc
a
r
e
,
a
c
or
ne
r
s
to
ne
of
s
o
c
ie
ta
l
w
e
ll
-
be
in
g,
is
unde
r
goi
ng
tr
a
ns
f
or
m
a
ti
ve
c
ha
nge
s
dr
iv
e
n
by
te
c
hnol
ogi
c
a
l
a
dva
nc
e
m
e
nt
s
.
A
m
ong
th
e
m
yr
ia
d
he
a
lt
h
c
ha
ll
e
nge
s
f
a
c
e
d
to
da
y,
di
a
be
te
s
e
m
e
r
ge
s
a
s
a
s
ig
ni
f
ic
a
nt
gl
oba
l
c
onc
e
r
n,
la
r
ge
ly
in
f
lu
e
nc
e
d
by
li
f
e
s
ty
le
c
h
a
nge
s
a
nd
in
c
r
e
a
s
in
g
pr
e
va
le
nc
e
.
T
hi
s
s
tu
dy
e
xpl
or
e
s
in
nova
ti
ve
te
c
hnol
ogi
c
a
l
s
ol
ut
io
ns
,
p
a
r
ti
c
ul
a
r
ly
da
ta
e
xt
r
a
c
ti
on
a
nd
f
uz
z
y
lo
gi
c
,
a
im
e
d
a
t
e
nha
nc
in
g
di
a
be
te
s
di
a
gnos
i
s
[
1]
. D
ia
be
te
s
i
s
c
ha
r
a
c
te
r
iz
e
d by poor
gl
uc
os
e
r
e
gul
a
ti
on, r
e
s
ul
ti
ng f
r
om
i
na
de
qua
te
i
ns
ul
in
pr
oduc
ti
on
or
in
e
f
f
e
c
ti
ve
in
s
ul
in
r
e
s
pon
s
e
,
l
e
a
di
ng
to
c
hr
oni
c
hi
gh
bl
ood
s
uga
r
l
e
ve
ls
(
hype
r
gl
yc
e
m
ia
)
.
W
hi
le
di
a
be
te
s
r
e
m
a
in
s
in
c
ur
a
bl
e
,
e
f
f
e
c
ti
ve
m
a
n
a
ge
m
e
nt
s
tr
a
te
gi
e
s
c
a
n
s
ig
ni
f
ic
a
nt
ly
im
pr
ove
pa
ti
e
nt
out
c
om
e
s
[
2]
.
T
he
e
s
c
a
la
ti
ng
in
c
id
e
nc
e
of
di
a
be
te
s
ne
c
e
s
s
it
a
te
s
ne
w
te
c
hnol
o
gi
c
a
l
in
te
r
ve
nt
io
ns
to
f
a
c
il
it
a
te
e
a
r
ly
de
te
c
ti
on
a
nd
tr
e
a
tm
e
nt
.
M
a
c
hi
ne
le
a
r
ni
ng
(
M
L
)
of
f
e
r
s
pr
om
is
in
g
o
ppor
tu
ni
ti
e
s
to
a
dva
nc
e
e
xi
s
ti
ng
he
a
lt
hc
a
r
e
te
c
hnol
ogi
e
s
w
it
hi
n
th
e
br
oa
de
r
c
ont
e
xt
of
th
e
f
our
th
in
dus
tr
ia
l
r
e
vol
ut
io
n,
w
hi
c
h
e
nc
om
pa
s
s
e
s
th
e
in
te
r
ne
t
of
th
in
gs
(
I
oT
)
,
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
(
A
I
)
,
da
ta
m
in
in
g,
a
nd
ne
ur
a
l
ne
twor
ks
.
D
e
s
pi
te
pr
e
vi
ous
r
e
s
e
a
r
c
h
e
f
f
or
ts
,
e
a
r
ly
di
a
be
te
s
de
te
c
ti
on
r
e
m
a
in
s
c
ha
ll
e
ngi
ng,
w
it
h
tr
a
di
ti
ona
l
m
e
th
ods
of
te
n
f
a
ll
in
g
s
hor
t
in
pr
ovi
di
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
2025
:
3681
-
3692
3682
c
om
pr
e
he
ns
iv
e
s
ol
ut
io
ns
.
T
hi
s
dr
iv
e
s
our
in
ve
s
ti
ga
ti
on
in
to
da
t
a
-
dr
iv
e
n
a
ppr
oa
c
he
s
,
f
oc
u
s
in
g
on
d
a
ta
m
in
in
g
a
nd f
uz
z
y l
ogi
c
, t
o e
nha
nc
e
di
a
gno
s
ti
c
a
c
c
ur
a
c
y
[
3]
.
T
hi
s
r
e
s
e
a
r
c
h
a
ddr
e
s
s
e
s
th
e
c
r
it
ic
a
l
is
s
ue
of
e
a
r
ly
di
a
be
te
s
di
a
gnos
is
th
r
ough
th
e
in
tr
oduc
ti
on
o
f
a
nove
l
c
lu
s
te
r
in
g
m
e
th
od,
e
va
lu
a
te
d
a
ga
in
s
t
e
s
ta
bl
is
he
d
a
lg
or
it
hm
s
.
T
he
s
tu
dy
c
ont
r
ib
ut
e
s
to
th
e
f
ie
ld
in
two
s
ig
ni
f
ic
a
nt
w
a
ys
:
‒
C
om
pr
e
he
ns
iv
e
a
na
ly
s
is
:
it
pr
e
s
e
nt
s
a
de
ta
il
e
d
c
om
pa
r
is
on
o
f
e
xi
s
ti
ng
c
lu
s
te
r
in
g
te
c
hni
que
s
us
e
d
in
di
a
be
te
s
pr
e
di
c
ti
on, highl
ig
ht
in
g t
he
ir
s
tr
e
ngt
hs
a
nd l
im
it
a
ti
ons
.
T
hi
s
c
om
pa
r
a
ti
ve
a
na
ly
s
is
w
il
l
s
e
r
ve
a
s
a
va
lu
a
bl
e
r
e
s
our
c
e
f
or
gui
di
ng f
ut
ur
e
r
e
s
e
a
r
c
h.
‒
I
nnova
ti
ve
c
lu
s
te
r
in
g
m
e
th
od
:
th
e
s
tu
dy
in
tr
oduc
e
s
a
n
e
w
c
lu
s
te
r
in
g
m
e
th
od
s
pe
c
if
ic
a
ll
y
de
s
ig
n
e
d
f
or
di
a
be
te
s
di
a
gno
s
is
, de
m
ons
tr
a
ti
ng ma
r
ke
d i
m
pr
ove
m
e
nt
s
i
n a
c
c
ur
a
c
y ove
r
c
onve
nt
io
na
l
te
c
hni
que
s
.
T
he
li
te
r
a
tu
r
e
r
e
vi
e
w
is
ope
ne
d
w
it
h
th
e
w
or
k
of
I
br
a
hi
m
e
t
al
.
[
4]
,
w
hi
c
h
pr
opos
e
s
a
ne
w
hybr
id
a
ppr
oa
c
h
th
a
t
c
om
bi
ne
s
a
ggl
om
e
r
a
ti
ve
hi
e
r
a
r
c
hi
c
a
l
c
lu
s
te
r
in
g
(
H
C
)
w
it
h
a
de
c
is
io
n
tr
e
e
c
la
s
s
if
ie
r
to
im
pr
ov
e
a
c
c
ur
a
c
y,
a
tt
a
in
in
g
a
n
80.8%
r
a
ti
ng
a
s
oppos
e
d
to
th
e
ir
c
ons
i
de
r
e
d
ty
pi
c
a
l
de
c
is
io
n
tr
e
e
c
la
s
s
if
ie
r
w
it
h
a
n
a
c
c
ur
a
c
y
r
a
ti
ng
of
76.9%
.
D
ong
e
t
al
.
[
5]
c
ont
r
ib
u
te
d
a
pr
oc
e
dur
e
us
in
g
f
uz
z
y
m
ode
l
in
H
C
th
a
t
id
e
nt
if
ie
s
c
lu
s
te
r
s
of
c
om
pl
e
x
a
nd
in
tr
ic
a
te
s
ha
pe
s
.
T
ha
t
a
lg
or
it
hm
r
e
ve
r
e
d
out
s
ta
ndi
ng
pe
r
f
or
m
a
nc
e
s
pe
c
if
ic
a
ll
y
to
hi
gh
-
di
m
e
ns
io
na
l
a
nd
la
r
ge
da
ta
s
e
ts
.
P
a
dm
a
ja
e
t
al
.
[
6]
ha
ve
t
a
ke
n
in
to
c
ons
id
e
r
a
ti
on
th
e
ta
s
k
of
id
e
nt
if
yi
n
g
hi
gh
-
qua
li
ty
c
lu
s
te
r
s
a
nd
m
a
d
e
a
de
e
p
a
na
ly
s
is
of
di
f
f
e
r
e
nt
a
lg
or
it
hm
s
f
or
c
lu
s
te
r
in
g.
G
hos
h
e
t
al
.
[
7]
c
ont
r
ib
ut
e
d t
o t
h
e
e
f
f
e
c
ti
ve
ne
s
s
of
t
h
e
a
g
gr
e
ga
ti
on ph
e
r
om
on
e
c
l
us
te
r
i
ng (
A
P
C
)
a
l
gor
it
hm
by
s
ho
w
in
g t
h
a
t
it
i
s
m
uc
h be
tt
e
r
c
o
nc
e
r
ni
ng
th
e
qua
li
ty
of
c
l
us
t
e
r
in
g a
nd
s
pe
e
d
of
pr
oc
e
s
s
i
ng
f
or
a
ll
th
e
da
t
a
s
e
t
s
t
a
ke
n
. B
a
gi
r
o
v
[
8]
pr
opos
e
d a
gl
oba
l
k
-
m
e
a
ns
(
K
M
)
a
lg
or
i
th
m
t
ha
t
s
how
e
d i
ts
e
f
f
e
c
ti
ve
ne
s
s
by be
in
g t
e
s
te
d on 14 da
ta
s
e
ts
us
in
g
num
e
r
ic
a
l
e
xpe
r
im
e
nt
s
,
th
ough
it
c
on
s
um
e
d
m
or
e
c
om
put
a
t
io
na
l
ti
m
e
.
F
in
a
ll
y,
N
it
hya
e
t
al
.
[
9]
ha
v
e
c
onduc
te
d
a
c
om
pa
r
a
ti
ve
s
tu
dy
on
H
C
,
d
e
ns
it
y
-
ba
s
e
d
s
p
a
ti
a
l
c
lu
s
te
r
in
g
of
a
ppl
ic
a
ti
ons
w
it
h
noi
s
e
(
D
B
S
C
A
N
)
,
a
nd
s
im
pl
e
K
M
s
c
he
m
e
a
nd
f
ound
th
a
t
th
e
KM
a
lg
or
it
hm
w
or
ks
be
s
t
on
th
e
d
ia
be
te
s
da
ta
s
e
t.
C
e
be
c
i
a
nd
Y
il
di
z
[
10]
a
ls
o
f
ound
th
e
K
M
a
lg
or
it
hm
to
be
f
a
s
te
r
in
e
xe
c
ut
io
n
a
s
c
om
pa
r
e
d
to
th
e
f
uz
z
y
C
-
m
e
a
ns
(
F
C
M
)
te
c
hni
que
f
or
a
ll
th
e
da
ta
s
e
ts
t
e
s
te
d,
i
nde
pe
nde
nt
of
th
e
ty
pe
of
pa
tt
e
r
n
in
th
e
b
a
s
e
da
ta
s
e
t.
T
hi
s
te
nde
n
c
y
to
w
a
r
ds
th
e
K
M
s
c
he
m
e
w
a
s
f
ur
th
e
r
c
onf
ir
m
e
d
by
B
ir
a
da
r
a
nd
M
uga
li
[
11
]
,
w
ho
a
ppl
ie
d
di
f
f
e
r
e
nt
to
ol
s
to
th
e
di
a
be
te
s
da
ta
s
e
t
.
Q
i
e
t
al
.
[
12]
d
e
s
ig
ne
d
a
n
a
ppr
oa
c
h
to
im
pr
ove
c
lu
s
te
r
in
g
by
c
hoos
in
g
in
it
ia
l
c
e
nt
e
r
s
w
it
h
gr
e
a
t
c
a
r
ts
,
th
e
r
e
by
s
ubs
ta
nt
ia
ll
y
i
m
pr
ovi
ng
th
e
li
ke
li
hood
of
obt
a
in
in
g
opt
im
a
l
lo
c
a
l
s
ol
ut
io
ns
.
S
a
r
a
va
na
na
th
a
n
a
nd
V
e
lm
ur
uga
n
[
13]
f
oc
us
e
d
on
a
na
ly
z
in
g
th
e
e
xe
c
ut
io
n
ti
m
e
of
bot
h
K
M
a
nd F
C
M
te
c
hni
que
s
, w
it
h K
M
c
on
s
is
te
nt
ly
out
pe
r
f
or
m
in
g F
C
M
i
n t
e
r
m
s
of
e
xe
c
ut
io
n t
im
e
.
I
n
s
um
m
a
r
y
,
th
e
r
e
s
e
a
r
c
h
p
r
o
vi
d
e
d
a
c
om
pr
e
h
e
n
s
iv
e
e
v
a
l
u
a
t
io
n
of
c
l
u
s
te
r
i
ng
a
lg
or
it
hm
p
e
r
f
or
m
a
n
c
e
,
u
n
d
e
r
s
c
o
r
i
ng
t
h
e
hy
br
id
m
od
e
l'
s
pr
om
i
s
i
n
g
ou
tc
o
m
e
s
a
n
d
th
e
s
u
s
ta
i
n
e
d
e
f
f
ic
i
e
n
c
y
of
t
he
K
M
t
e
c
hn
iq
u
e
,
m
a
ki
ng
i
t
t
h
e
p
r
e
f
e
r
r
e
d
c
h
oi
c
e
f
o
r
l
a
r
g
e
d
a
t
a
s
e
t
s
.
A
d
di
ti
o
n
a
ll
y
,
it
e
m
p
h
a
s
iz
e
d
th
e
i
m
p
or
t
a
n
c
e
o
f
i
d
e
n
ti
f
yi
n
g
h
ig
h
-
qu
a
li
ty
c
l
u
s
t
e
r
s
a
s
a
m
e
a
n
s
t
o
a
u
gm
e
n
t
c
lu
s
t
e
r
i
n
g a
l
g
or
i
t
hm
e
f
f
e
c
t
iv
e
ne
s
s
.
O
r
a
b
i
e
t
al
.
[
1
4]
i
nt
r
o
du
c
e
s
a
n e
a
r
ly
p
r
e
d
ic
ti
v
e
s
y
s
t
e
m
f
or
d
i
a
b
e
te
s
m
e
ll
it
u
s
b
y
i
n
te
gr
a
ti
ng
d
a
t
a
m
i
ni
ng
t
e
c
hn
iq
ue
s
,
de
m
on
s
tr
a
t
in
g
im
pr
ov
e
d
pr
e
di
c
ti
on
a
c
c
ur
a
c
y
t
hr
o
ug
h
t
a
i
l
or
e
d
pr
e
pr
o
c
e
s
s
in
g
a
n
d
c
l
a
s
s
i
f
i
c
a
t
io
n
m
e
th
o
d
s
.
P
a
ti
l
e
t
al
.
[
15
]
p
r
o
po
s
e
a
hy
br
i
d
pr
e
d
i
c
t
io
n
m
od
e
l
f
or
t
y
p
e
-
2
d
i
a
b
e
te
s
t
h
a
t
c
om
bi
n
e
s
d
e
c
i
s
i
o
n
t
r
e
e
s
a
n
d
a
d
a
pt
iv
e
n
e
ur
o
-
f
u
z
z
y
i
nf
e
r
e
nc
e
s
y
s
te
m
s
(
A
N
F
I
S
)
,
yi
e
ld
in
g
s
u
p
e
r
io
r
p
e
r
f
o
r
m
a
n
c
e
c
om
p
a
r
e
d
t
o
s
t
a
n
da
lo
n
e
m
o
d
e
l
s
.
Z
ha
o
e
t
a
l
.
[
16]
c
o
nt
r
i
bu
te
t
o
t
h
e
e
v
a
lu
a
ti
on
of
c
lu
s
t
e
r
i
n
g
q
u
a
li
ty
by
pr
e
s
e
n
ti
n
g
a
s
um
-
of
-
s
q
u
a
r
e
s
-
b
a
s
e
d
c
lu
s
te
r
v
a
li
di
ty
in
d
e
x
w
i
th
s
ig
ni
f
i
c
a
n
c
e
a
na
ly
s
i
s
,
e
n
a
b
li
n
g
b
e
t
t
e
r
a
s
s
e
s
s
m
e
nt
a
nd
s
e
l
e
c
t
io
n
o
f
c
l
u
s
t
e
r
i
ng r
e
s
u
lt
s
.
B
a
h
m
a
n
i
e
t
al
.
[
17]
a
dd
r
e
s
s
t
h
e
s
c
a
l
a
b
il
it
y c
ha
ll
e
ng
e
s
i
n c
l
u
s
t
e
r
in
g
l
a
r
g
e
d
a
ta
s
e
t
s
th
r
o
u
gh
a
n
o
pt
im
i
z
e
d
k
-
m
e
a
n
s
+
+
a
l
g
or
i
th
m
,
a
c
h
ie
v
i
ng
f
a
s
t
e
r
e
x
e
c
ut
io
n
ti
m
e
s
w
hi
l
e
m
a
in
t
a
i
ni
n
g
h
i
gh
c
l
u
s
t
e
r
in
g
a
c
c
u
r
a
c
y
.
K
a
r
e
g
ow
d
a
e
t
al
.
[
18
]
e
xp
lo
r
e
a
c
a
s
c
a
di
n
g
a
p
pr
o
a
c
h
t
h
a
t
in
t
e
g
r
a
t
e
s
k
-
m
e
a
n
s
c
lu
s
t
e
r
i
n
g
w
i
th
k
-
ne
a
r
e
s
t
n
e
ig
hb
or
c
l
a
s
s
if
ic
a
ti
on
f
or
c
a
t
e
go
r
i
z
in
g
d
ia
b
e
t
i
c
p
a
ti
e
nt
s
,
h
ig
hl
ig
h
ti
ng
i
m
pr
o
v
e
d
c
la
s
s
if
i
c
a
ti
o
n
p
r
e
c
i
s
io
n
t
hr
ou
g
h
a
t
w
o
-
s
t
a
g
e
p
r
o
c
e
s
s
i
n
g
f
r
a
m
e
w
or
k
.
T
h
a
kk
a
r
e
t
al
.
[
19
]
c
o
m
p
a
r
e
d
da
t
a
m
in
i
ng
a
nd
f
u
z
z
y
l
og
i
c
t
e
c
hn
iq
u
e
s
f
o
r
di
a
be
t
e
s
pr
og
no
s
i
s
,
no
ti
ng
t
h
a
t
d
a
t
a
m
i
ni
ng
m
e
t
h
od
s
l
ik
e
d
e
c
i
s
io
n
tr
e
e
s
a
nd
s
up
po
r
t
v
e
c
t
or
m
a
c
hi
n
e
s
(
S
V
M
)
pr
ov
id
e
h
ig
h
a
c
c
u
r
a
c
y
b
ut
l
im
it
e
d
i
nt
e
r
pr
e
t
a
b
il
i
t
y.
I
n
c
o
nt
r
a
s
t
,
f
u
z
z
y
l
og
i
c
h
a
n
dl
e
s
u
n
c
e
r
t
a
in
ty
w
e
l
l
a
nd
o
f
f
e
r
s
t
r
a
n
s
pa
r
e
n
t,
r
ul
e
-
b
a
s
e
d
r
e
a
s
o
ni
n
g
a
li
gn
e
d
w
it
h
c
li
ni
c
a
l
pr
a
c
t
i
c
e
.
T
he
s
tr
uc
tu
r
e
of
th
is
pa
pe
r
is
a
s
f
ol
lo
w
s
:
s
e
c
ti
on
2
de
ta
il
s
th
e
m
e
th
odol
ogy,
out
li
ni
ng
th
e
c
om
pa
r
a
ti
ve
s
tu
dy
of
e
xi
s
ti
ng
pr
oc
e
dur
e
s
a
nd
th
e
pr
opos
e
d
n
e
w
m
e
th
od.
S
e
c
ti
on
3
pr
e
s
e
nt
s
th
e
c
om
pa
r
is
on
r
e
s
ul
ts
,
f
ol
lo
w
e
d
by
a
th
or
ough
di
s
c
us
s
io
n
of
th
e
f
in
di
ngs
.
F
in
a
ll
y,
s
e
c
ti
on
4
s
um
m
a
r
iz
e
s
ke
y
in
s
ig
ht
s
,
e
m
pha
s
iz
in
g
th
e
im
pl
ic
a
ti
ons
a
nd
c
ont
r
ib
ut
io
ns
of
th
e
nove
l
m
e
th
od
w
hi
le
s
ugge
s
ti
ng
a
ve
nu
e
s
f
or
f
ur
th
e
r
r
e
s
e
a
r
c
h a
nd de
v
e
lo
pm
e
nt
i
n di
a
be
te
s
di
a
gno
s
is
t
e
c
hnol
ogy.
2.
M
E
T
H
O
D
T
he
obj
e
c
ti
ve
of
th
is
r
e
s
e
a
r
c
h
is
to
pr
opos
e
a
de
c
i
s
io
n
-
m
a
ki
ng
c
lu
s
te
r
in
g
a
ppr
oa
c
h
f
or
ha
ndl
in
g
di
a
be
te
s
-
r
e
la
te
d
a
tt
r
ib
ut
e
s
in
th
e
P
im
a
I
ndi
a
ns
di
a
be
te
s
da
ta
s
e
t
(
P
I
D
D
)
.
T
hi
s
s
e
c
ti
on
out
li
ne
s
th
e
s
ys
te
m
a
ti
c
m
e
th
odol
ogy
e
m
pl
oye
d
to
c
la
s
s
if
y
di
a
be
te
s
a
tt
r
ib
ut
e
s
f
or
th
e
in
it
ia
l
de
te
c
ti
on
a
nd
pr
e
di
c
ti
on
of
d
ia
be
te
s
,
de
ta
il
in
g t
he
e
xpe
r
im
e
nt
a
l
pr
oc
e
dur
e
s
f
ol
lo
w
e
d t
o e
ns
ur
e
r
e
pr
o
duc
ib
il
it
y.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
in
g di
abe
te
s
pr
e
di
c
ti
on:
unv
e
il
in
g pati
e
nt
s
ubgr
oup
s
t
hr
ough c
lu
s
te
r
in
g
(
R
it
a G
anguly
)
3683
2.1. Dat
a p
r
e
-
p
r
oc
e
s
s
in
g
T
he
r
e
a
r
e
t
w
o m
e
th
ods
f
or
da
ta
pr
e
pr
oc
e
s
s
in
g:
2.1.1. Dat
a e
xt
r
ac
t
io
n
an
d
c
le
an
in
g
T
he
P
I
D
D
w
a
s
e
xt
r
a
c
te
d
a
nd
e
xa
m
in
e
d
f
or
qua
li
ty
.
M
is
s
in
g
va
lu
e
s
w
e
r
e
a
ddr
e
s
s
e
d
us
in
g
s
e
v
e
r
a
l
im
put
a
ti
on
te
c
hni
que
s
.
T
he
s
e
in
c
lu
de
d
m
e
a
n
or
m
e
di
a
n
im
put
a
ti
on,
f
or
w
a
r
d
a
nd
ba
c
kw
a
r
d
f
i
ll
,
m
ul
ti
pl
e
im
put
a
ti
on, a
nd mode
l
-
ba
s
e
d i
m
put
a
ti
on.
2.1.2. Nor
m
al
iz
at
io
n
an
d
s
c
al
in
g
T
o
pr
e
ve
nt
bi
a
s
to
w
a
r
d
a
tt
r
ib
ut
e
s
w
it
h
la
r
ge
r
va
lu
e
s
,
nor
m
a
li
z
a
ti
on
w
a
s
p
e
r
f
or
m
e
d
on
a
ll
da
ta
s
e
t
a
tt
r
ib
ut
e
s
. T
he
nor
m
a
li
z
a
ti
on e
qua
ti
on u
s
e
d w
a
s
:
X
n
o
r
m
=
−
−
w
he
r
e
X
is
th
e
or
ig
in
a
l
a
tt
r
ib
ut
e
va
lu
e
;
X
m
in
a
nd
X
m
a
x
a
r
e
t
he
lo
w
e
s
t
a
nd
e
xt
r
e
m
e
v
a
lu
e
s
of
th
e
a
tt
r
ib
ut
e
r
e
s
pe
c
ti
ve
ly
;
a
nd X
nor
m
i
s
t
he
nor
m
a
li
z
e
d a
tt
r
ib
ut
e
va
lu
e
.
2.2. Com
p
ar
at
iv
e
an
al
ys
is
o
f
c
lu
s
t
e
r
in
g al
gor
it
h
m
s
T
hr
e
e
popula
r
c
lu
s
te
r
in
g
a
lg
or
it
hm
s
a
r
e
s
e
le
c
te
d
f
or
a
na
ly
s
is
:
K
M
,
F
C
M
,
a
nd
H
C
.
E
a
c
h
c
lu
s
te
r
in
g
a
lg
or
it
hm
is
a
ppl
ie
d
to
th
e
pr
e
-
pr
oc
e
s
s
e
d
da
ta
s
e
t
to
c
r
e
a
te
c
lu
s
te
r
s
of
di
a
be
te
s
-
r
e
la
te
d
a
tt
r
ib
ut
e
s
.
T
he
f
unda
m
e
nt
a
l
pr
in
c
ip
le
s
of
e
a
c
h c
lu
s
te
r
in
g a
lg
or
it
hm
a
r
e
s
tu
di
e
d
a
nd unde
r
s
to
od.
2.2.1. K
-
m
e
an
s
c
lu
s
t
e
r
in
g
K
-
m
e
a
ns
s
e
e
k
s
to
gr
oup
da
ta
s
o
th
a
t
poi
nt
s
in
th
e
s
a
m
e
c
lu
s
te
r
a
r
e
a
s
c
lo
s
e
a
s
pos
s
ib
le
to
th
e
ir
c
lu
s
te
r
’
s
c
e
nt
e
r
.
O
bj
e
c
ti
ve
f
unc
ti
on:
=
∑
∑
|
|
−
|
|
n
i
=
1
=
1
2
w
he
r
e
th
e
num
be
r
of
c
lu
s
te
r
s
is
r
e
pr
e
s
e
nt
e
d
by
K
;
da
ta
poi
nt
s
in
th
e
it
h
c
lu
s
te
r
is
n
i
;
jt
h
da
ta
poi
nt
in
th
e
i
th
c
lu
s
te
r
i
s
X
j
[
i
]
;
a
nd c
e
nt
r
oi
d of
t
he
i
th
gr
oup is
µ
i
.
2.2.2. F
u
z
z
y C
-
m
e
an
s
c
lu
s
t
e
r
in
g
F
uz
z
y C
-
m
e
a
ns
a
ll
ow
s
pa
r
ti
a
l
m
e
m
be
r
s
hi
p of
da
ta
poi
nt
s
i
n m
u
lt
ip
le
c
lu
s
te
r
s
.
O
bj
e
c
ti
ve
f
unc
ti
on:
=
∑
∑
|
|
−
|
|
2
=
1
=
1
w
he
r
e
n
is
th
e
da
ta
poi
nt
s
;
th
e
num
be
r
of
c
lu
s
te
r
s
is
k;
th
e
m
e
m
be
r
s
hi
p
de
gr
e
e
of
X
i
in
th
e
jt
h
c
lu
s
te
r
is
U
ij
;
th
e
f
uz
z
in
e
s
s
e
xpone
nt
i
s
m
;
it
h da
ta
poi
nt
i
s
X
i
;
M
id
-
poi
nt
of
t
he
j
th
c
lu
s
te
r
i
s
C
j
.
2.2.3.
H
ie
r
ar
c
h
ic
al
c
lu
s
t
e
r
in
g
HC
f
or
m
s
a
t
r
e
e
-
li
ke
s
tr
uc
tu
r
e
(
de
ndr
ogr
a
m
)
of
ne
s
te
d c
lu
s
te
r
s
.
L
in
ka
ge
f
unc
ti
on:
d
(
A
,
B
)
=
√
2
|
|
|
|
|
|
+
|
|
|
|
−
|
|
2
w
he
r
e
A
a
nd
B
a
r
e
two
c
lu
s
te
r
s
;
|A
|
a
nd
|B
|
a
r
e
th
e
s
iz
e
s
of
c
lu
s
te
r
s
A
a
nd
B
s
e
p
a
r
a
te
ly
;
a
nd
a
r
e
th
e
c
e
nt
r
oi
ds
of
c
lu
s
te
r
s
A
a
nd
B
s
e
p
a
r
a
te
ly
.
2.3. Valu
at
io
n
m
e
t
r
ic
s
N
um
e
r
ous
ke
y m
e
tr
ic
s
a
r
e
ut
il
iz
e
d
to
c
a
l
c
ul
a
te
t
he
c
lu
s
te
r
in
g a
lg
or
it
hm
’
s
pe
r
f
or
m
a
nc
e
:
‒
S
il
houe
tt
e
s
c
or
e
de
te
r
m
in
e
s
t
he
f
ir
m
ne
s
s
a
nd c
lu
s
te
r
s
e
pa
r
a
ti
on.
=
−
(
,
)
w
he
r
e
r
e
gul
a
r
in
tr
a
-
c
lu
s
te
r
di
s
ta
nc
e
i
s
a
;
a
nd t
ypi
c
a
l
a
dj
a
c
e
nt
-
c
l
us
te
r
di
s
ta
nc
e
i
s
b.
‒
A
dj
us
te
d
R
a
nd
in
de
x
(
A
R
I
)
s
c
or
e
c
om
pa
r
e
s
th
e
s
im
il
a
r
it
y
be
twe
e
n
tr
ue
c
la
s
s
la
be
l
s
a
nd
c
lu
s
te
r
a
s
s
ig
nm
e
nt
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
2025
:
3681
-
3692
3684
=
−
|
|
|
|
−
|
|
w
he
r
e
th
is
f
oc
us
e
s
on
th
e
c
onc
e
pt
of
|
|
a
s
a
m
e
a
s
ur
e
of
how
w
e
ll
;
a
nd
th
e
R
a
nd
in
de
x
(
RI
)
is
e
xpe
c
te
d t
o pe
r
f
or
m
on a
ve
r
a
ge
.
‒
N
or
m
a
li
z
e
d m
ut
ua
l
in
f
or
m
a
ti
on (
N
M
I
)
s
c
or
e
:
th
e
N
M
I
qua
nt
if
i
e
s
t
he
a
gr
e
e
m
e
nt
be
tw
e
e
n t
r
ue
c
la
s
s
l
a
be
ls
a
nd c
lu
s
te
r
a
s
s
ig
nm
e
nt
s
, a
c
c
ount
in
g f
or
e
nt
r
opy.
(
,
)
=
2
(
;
)
(
)
+
(
)
w
he
r
e
th
e
M
I
a
m
id
c
lu
s
te
r
s
U
a
nd
V
is
I
(
U
;
V
)
;
a
nd
th
e
r
a
ndomne
s
s
of
c
lu
s
te
r
s
U
a
nd
V
a
r
e
H
(
U
)
a
nd
H
(
V
)
r
e
s
pe
c
ti
ve
ly
.
‒
D
a
vi
e
s
-
B
oul
di
n
in
de
x (
D
B
I
)
s
c
or
e
:
de
f
in
e
s
t
he
c
lu
s
te
r
s
upe
r
io
r
i
ty
ba
s
e
d on the
di
s
ta
n
c
e
be
twe
e
n c
lu
s
te
r
s
.
=
1
∑
≠
=
1
(
+
)
(
,
)
w
he
r
e
S
i
a
nd
S
j
is
th
e
ty
pi
c
a
l
di
s
ta
nc
e
be
twe
e
n
e
a
c
h
poi
nt
in
c
l
us
te
r
i
a
nd
th
e
c
e
nt
r
oi
d
C
i
;
a
nd
d(
C
i
,
C
j
)
is
th
e
c
e
nt
r
oi
d di
s
ta
nc
e
be
tw
e
e
n C
i
a
nd C
j
.
T
he
s
e
e
va
lu
a
ti
on
m
e
a
s
ur
e
s
h
e
lp
r
e
s
e
a
r
c
h
e
r
s
de
te
r
m
in
e
how
w
e
ll
th
e
c
lu
s
te
r
s
r
e
pr
e
s
e
nt
th
e
unde
r
ly
in
g
da
ta
pa
tt
e
r
ns
.
B
y
c
om
pa
r
in
g
th
e
r
e
s
ul
ts
a
c
r
os
s
di
f
f
e
r
e
nt
m
e
th
ods
,
th
e
y
c
a
n
s
e
le
c
t
th
e
c
lu
s
te
r
in
g
a
ppr
oa
c
h
th
a
t
pr
ovi
de
s
t
he
m
os
t
a
c
c
ur
a
te
a
nd me
a
ni
ngf
ul
gr
oupi
ng, a
s
de
pi
c
te
d i
n F
ig
ur
e
1.
F
ig
ur
e
1. E
va
lu
a
ti
on me
tr
ic
s
2.4. I
n
n
ovat
iv
e
c
lu
s
t
e
r
in
g m
e
t
h
od
f
or
m
u
la
t
io
n
A
n
ov
e
l
c
l
u
s
te
r
in
g m
e
t
hod
e
m
e
r
g
e
s
f
r
o
m
t
he
c
om
p
a
r
a
ti
ve
a
na
ly
s
i
s
a
n
d
e
v
a
lu
a
t
io
n
m
e
tr
i
c
s
.
T
hi
s
m
e
th
od
is
d
e
s
ig
n
e
d
to
opt
im
i
z
e
th
e
c
l
u
s
t
e
r
in
g
of
di
a
be
t
e
s
a
tt
r
i
but
e
s
a
nd
e
nh
a
n
c
e
th
e
a
c
c
ur
a
c
y
of
di
s
e
a
s
e
pr
e
d
ic
ti
o
n
.
I
t
pot
e
nt
ia
ll
y
d
oe
s
s
o b
y
in
t
e
gr
a
t
in
g
e
le
m
e
nt
s
f
r
om
e
xi
s
ti
ng
a
l
gor
i
th
m
s
or
in
tr
od
uc
in
g
e
nt
ir
e
l
y
ne
w
a
p
pr
o
a
c
he
s
.
2.5. As
s
e
s
s
m
e
n
t
an
d
ou
t
c
om
e
s
T
he
ne
w
ly
de
ve
lo
p
e
d
c
lu
s
te
r
in
g
m
e
th
od
is
im
pl
e
m
e
nt
e
d
a
nd
be
nc
hm
a
r
ke
d
a
ga
in
s
t
K
M
,
F
C
M
,
a
nd
H
C
us
in
g
th
e
P
I
D
D
.
I
ts
pe
r
f
or
m
a
nc
e
is
a
s
s
e
s
s
e
d
u
s
in
g
s
ta
nda
r
d
e
va
lu
a
ti
on
m
e
tr
ic
s
.
T
h
e
s
e
m
e
tr
ic
s
a
r
e
us
e
d
to
de
m
ons
tr
a
te
t
he
e
f
f
e
c
ti
ve
ne
s
s
of
t
he
pr
opos
e
d m
e
th
od.
2.6. I
m
p
li
c
at
io
n
s
an
d
ap
p
li
c
at
io
n
s
T
he
s
tu
dy
c
on
s
id
e
r
s
th
e
po
s
s
ib
il
it
ie
s
f
or
th
e
e
a
r
ly
a
nd
c
or
r
e
c
t
di
a
gnos
is
of
di
a
be
te
s
by
im
pr
ovi
ng
c
lu
s
te
r
in
g
te
c
hni
que
s
f
or
a
tt
r
ib
ut
e
s
r
e
la
te
d
to
di
a
be
te
s
.
I
t
in
ve
s
ti
ga
te
s
how
in
nova
ti
ve
a
lg
or
it
hm
s
of
c
lu
s
te
r
in
g
c
oul
d
be
put
in
to
a
c
ti
on
to
a
c
hi
e
ve
opt
im
a
li
ty
.
T
h
e
s
e
im
pr
ove
m
e
nt
s
a
im
to
s
uppor
t
s
tr
a
te
gi
e
s
f
or
e
f
f
e
c
ti
ve
he
a
lt
h de
c
is
io
n
-
m
a
ki
ng a
nd dis
e
a
s
e
m
a
na
ge
m
e
nt
.
2.7. Hi
gh
li
gh
t
e
d
i
m
p
ac
t
T
he
c
ha
ll
e
ng
e
of
di
a
be
te
s
is
e
m
e
r
gi
ng
a
nd
is
be
in
g
a
ddr
e
s
s
e
d
th
r
ough
da
ta
-
dr
iv
e
n
a
na
ly
ti
c
s
in
th
e
r
e
s
e
a
r
c
h.
I
t
h
a
s
c
om
e
up
w
it
h
a
ne
w
te
c
hni
que
of
c
lu
s
te
r
in
g
a
nd
ha
s
c
om
pa
r
e
d
it
w
it
h
ot
he
r
m
e
th
ods
to
id
e
nt
if
y
th
e
ga
ps
in
th
e
e
a
r
ly
di
a
gnos
is
of
di
a
be
te
s
.
E
xt
e
ns
iv
e
e
xpe
r
im
e
nt
s
a
r
e
be
in
g
c
onduc
te
d
to
c
om
e
up
w
it
h
a
di
a
be
te
s
a
tt
r
ib
ut
e
c
lu
s
te
r
in
g
te
c
hni
que
b
e
tt
e
r
th
a
n
th
e
one
s
e
xi
s
ti
ng
[
20]
.
T
hr
ough
th
is
s
tr
uc
tu
r
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
in
g di
abe
te
s
pr
e
di
c
ti
on:
unv
e
il
in
g pati
e
nt
s
ubgr
oup
s
t
hr
ough c
lu
s
te
r
in
g
(
R
it
a G
anguly
)
3685
m
e
th
odol
ogy,
th
e
r
e
s
e
a
r
c
h
a
im
s
to
a
ddr
e
s
s
th
e
c
ha
ll
e
ng
e
of
di
a
be
te
s
e
a
r
ly
de
te
c
ti
on
a
nd
m
a
na
ge
m
e
nt
,
de
m
ons
tr
a
ti
ng
th
e
pow
e
r
of
da
ta
-
dr
iv
e
n
a
ppr
oa
c
he
s
in
he
a
lt
hc
a
r
e
a
na
ly
ti
c
s
.
B
y
f
ol
lo
w
in
g
th
e
s
e
s
te
ps
,
f
ut
ur
e
r
e
s
e
a
r
c
he
r
s
c
a
n r
e
pl
ic
a
te
t
he
e
xp
e
r
im
e
nt
s
a
nd buil
d upon the f
in
di
ngs
pr
e
s
e
nt
e
d i
n F
ig
ur
e
2.
F
ig
ur
e
2. P
r
opos
e
d f
r
a
m
e
w
or
k f
or
c
lu
s
te
r
in
g
3.
R
E
S
U
L
T
S
A
N
D
D
I
S
C
U
S
S
I
O
N
3.1.
I
d
e
n
t
if
yi
n
g gap
s
i
n
p
r
e
vi
ou
s
r
e
s
e
a
r
c
h
T
hi
s
s
tu
dy
in
v
e
s
ti
ga
t
e
s
th
e
pe
r
f
or
m
a
nc
e
of
va
r
i
ous
c
lu
s
t
e
r
in
g
a
lg
or
it
hm
s
—
KM
,
F
C
M
,
a
nd
HC
—
in
th
e
c
o
nt
e
x
t
of
e
a
r
ly
di
a
be
t
e
s
di
a
gno
s
is
.
W
hi
l
e
pr
io
r
r
e
s
e
a
r
c
h
h
a
s
e
xpl
or
e
d
th
e
e
f
f
ic
a
c
y
of
c
l
us
t
e
r
in
g
t
e
c
h
ni
qu
e
s
in
he
a
lt
hc
a
r
e
da
t
a
,
m
a
n
y
ha
v
e
not
e
x
pl
ic
it
ly
a
ddr
e
s
s
e
d
how
th
e
s
e
m
e
t
hod
s
c
a
n
be
opt
i
m
iz
e
d
f
or
da
ta
s
e
ts
w
i
th
im
ba
la
nc
e
d c
l
a
s
s
e
s
a
nd
m
is
s
in
g
va
lu
e
s
.
T
hi
s
ga
p i
s
pa
r
t
ic
ul
a
r
ly
r
e
le
v
a
nt
i
n
pr
e
di
c
ti
ng
di
a
be
t
e
s
out
c
om
e
s
.
3.2.
S
u
m
m
ar
iz
in
g k
e
y f
in
d
in
gs
I
n
th
is
r
e
s
e
a
r
c
h
w
or
k
f
in
di
ngs
in
di
c
a
te
th
a
t
KM
c
lu
s
te
r
in
g
pr
oduc
e
d
th
e
hi
ghe
s
t
a
c
c
ur
a
c
y
m
e
tr
ic
s
,
a
c
hi
e
vi
ng
a
pe
r
f
e
c
t
s
c
or
e
a
c
r
os
s
m
ul
ti
pl
e
e
va
lu
a
ti
on
pa
r
a
m
e
t
e
r
s
.
I
n
c
ont
r
a
s
t,
F
C
M
e
xhi
bi
te
d
a
s
ig
ni
f
ic
a
nt
ly
lo
w
e
r
pe
r
f
or
m
a
nc
e
,
pa
r
ti
c
ul
a
r
ly
in
s
e
n
s
it
iv
it
y
a
nd
s
p
e
c
if
ic
it
y,
s
ugge
s
ti
ng
it
s
li
m
it
a
ti
ons
in
c
le
a
r
bounda
r
y
de
li
ne
a
ti
on
a
m
ong
ove
r
la
ppi
ng
c
lu
s
te
r
s
.
HC
de
m
ons
tr
a
te
d
m
ode
r
a
te
e
f
f
e
c
ti
ve
ne
s
s
but
s
tr
uggl
e
d
w
it
h
la
r
ge
da
ta
s
e
ts
du
e
t
o i
ts
c
om
put
a
ti
ona
l
in
te
ns
it
y.
‒
KM
:
c
e
nt
r
oi
d a
na
ly
s
i
s
.
‒
F
C
M
:
m
e
m
be
r
s
hi
p va
lu
e
a
n
a
ly
s
is
.
‒
V
is
ua
li
z
a
ti
on:
vi
s
ua
li
z
in
g c
lu
s
te
r
s
i
n r
e
duc
e
d
-
di
m
e
ns
io
na
l
s
pa
c
e
.
‒
F
e
a
tu
r
e
i
m
por
ta
nc
e
:
f
e
a
tu
r
e
s
w
it
h l
a
r
ge
r
pe
r
f
or
m
a
nc
e
c
ha
nge
s
u
pon pe
r
m
ut
a
ti
on a
r
e
m
or
e
i
nf
lu
e
nt
ia
l.
‒
C
ont
in
uous
im
pr
ove
m
e
nt
:
to
e
nha
nc
e
a
lg
or
it
hm
a
da
pt
a
bi
li
ty
,
r
e
tr
a
in
w
it
h
ne
w
da
ta
,
us
e
in
c
r
e
m
e
nt
a
l
le
a
r
ni
ng t
e
c
hni
que
s
, a
nd moni
to
r
da
ta
di
s
tr
ib
ut
io
n f
or
pe
r
io
di
c
m
ode
l
r
e
tr
a
in
in
g.
U
nve
il
in
g
da
ta
'
s
hi
dde
n
p
a
tt
e
r
ns
de
m
a
nd
s
th
e
p
e
r
f
e
c
t
c
lu
s
te
r
in
g
f
it
.
R
e
s
e
a
r
c
h
e
r
s
m
a
tc
h
d
a
ta
s
e
t
tr
a
it
s
a
nd
a
na
ly
s
is
goa
l
s
to
th
e
id
e
a
l
a
lg
or
it
hm
.
C
lu
s
te
r
di
s
ta
nc
e
s
a
nd
e
s
ta
bl
is
h
e
d
m
e
tr
ic
s
gui
de
th
e
c
hoi
c
e
,
a
lo
ngs
id
e
dom
a
in
knowle
dge
.
V
is
u
a
li
z
in
g
th
e
c
lu
s
te
r
s
pr
ovi
de
s
a
f
in
a
l
th
um
bs
-
up
on
th
e
ir
qua
li
ty
a
nd
e
f
f
e
c
ti
ve
ne
s
s
.
T
he
d
a
ta
ba
s
e
de
s
c
r
ip
ti
on
a
nd
th
e
c
lu
s
te
r
in
g
m
e
c
h
a
ni
s
m
a
r
e
c
o
ve
r
e
d
in
th
is
s
e
c
ti
on.
K
M
,
F
C
M
,
a
nd
H
C
a
lg
or
it
hm
s
a
r
e
us
e
d
f
or
th
e
a
na
ly
s
is
a
nd
to
obt
a
in
th
e
hi
ghe
s
t
a
c
c
ur
a
c
y
w
it
h
th
e
pr
e
di
c
te
d
m
ode
l.
T
he
pr
opos
e
d
m
e
th
od
is
im
pl
e
m
e
nt
e
d
us
in
g
P
yt
hon
ve
r
s
io
n
3.
11.3,
I
nt
e
l
(
R
)
C
or
e
(
T
M
)
i.
3
7020
you
C
P
U
@
2. 30 G
H
z
w
it
h 8 G
B
R
A
M
.
3.2.1.
D
at
as
e
t
H
e
r
e
,
th
e
P
I
D
D
is
us
e
d,
w
hi
c
h
c
ont
a
in
s
in
f
or
m
a
ti
on
on
768
pa
ti
e
nt
s
.
A
m
ong
th
e
768
pa
ti
e
nt
s
,
onl
y
268
pa
ti
e
nt
s
(
34.9%
)
w
e
r
e
c
la
s
s
if
ie
d
a
s
ha
vi
ng
pos
it
iv
e
di
a
be
t
e
s
.
T
he
da
t
a
s
e
t
h
a
s
8
a
tt
r
ib
ut
e
s
w
it
h
one
c
l
a
s
s
a
tt
r
ib
ut
e
w
he
r
e
th
e
c
la
s
s
v
a
lu
e
be
lo
ng
s
to
0
a
nd
1.
T
a
bl
e
1
p
r
e
s
e
nt
s
th
e
a
tt
r
ib
ut
e
s
a
nd
th
e
ir
c
or
r
e
s
ponding
num
be
r
of
m
is
s
in
g va
lu
e
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
2025
:
3681
-
3692
3686
T
a
bl
e
1
.
C
onf
ig
ur
a
ti
on of
d
a
ta
s
e
t
A
t
t
r
i
but
e
s
T
ot
a
l
no of
m
i
s
s
i
ng va
l
ue
P
r
e
g
0
P
l
a
s
5
P
r
e
s
28
S
ki
n
192
I
ns
u
140
M
a
s
s
11
P
e
di
0
A
ge
0
C
l
a
s
s
0
3.3. I
n
t
e
r
p
r
e
t
in
g r
e
s
u
lt
s
:
c
om
p
ar
i
s
on
w
it
h
ot
h
e
r
s
t
u
d
ie
s
T
he
a
s
s
e
s
s
m
e
nt
of
pe
r
f
or
m
a
nc
e
in
c
lu
de
s
th
e
c
om
put
a
ti
on
o
f
s
e
ve
r
a
l
m
e
tr
ic
s
s
uc
h
a
s
a
c
c
ur
a
c
y,
s
e
ns
it
iv
it
y,
pr
e
c
is
io
n,
s
pe
c
if
ic
it
y,
F
1
-
s
c
or
e
,
a
nd
e
r
r
or
r
a
te
.
A
c
c
ur
a
c
y
is
de
te
r
m
in
e
d
by
m
e
a
s
ur
in
g
th
e
a
m
ount
of
a
ppr
opr
ia
te
ly
pr
ophe
s
ie
d i
ll
us
tr
a
ti
ons
out
of
t
he
t
ot
a
l
in
s
ta
nc
e
s
.
‒
A
c
c
ur
a
c
y:
t
he
r
a
ti
o of
a
ll
pr
e
c
is
e
ly
f
or
e
c
a
s
te
d
s
a
m
pl
e
s
t
o t
he
t
ot
a
l
num
be
r
of
s
a
m
pl
e
s
;
a
s
e
xpr
e
s
s
e
d i
n (
1)
=
+
+
+
+
(
1)
‒
S
e
ns
it
iv
it
y:
c
a
te
gor
iz
a
ti
on of
pos
it
iv
e
s
a
m
pl
e
s
.
T
hi
s
i
s
m
a
th
e
m
a
ti
c
a
ll
y e
xpr
e
s
s
e
d i
n (
2)
=
(
+
)
(
2)
‒
P
r
e
c
is
io
n:
th
e
pr
opor
ti
on
of
th
e
num
be
r
of
pr
e
c
is
e
ly
f
or
e
c
a
s
te
d
in
s
ta
nc
e
s
to
th
e
to
ta
l
num
be
r
of
pos
it
iv
e
s
a
m
pl
e
s
.
T
hi
s
i
s
m
a
th
e
m
a
ti
c
a
ll
y e
xpr
e
s
s
e
d i
n (
3)
=
(
+
)
(
3)
‒
S
pe
c
if
ic
it
y:
c
a
te
gor
iz
e
s
ne
ga
ti
ve
s
a
m
pl
e
s
.
T
hi
s
i
s
m
a
th
e
m
a
ti
c
a
ll
y e
xpr
e
s
s
e
d i
n (
4)
=
(
+
)
(
4)
‒
F1
-
s
c
or
e
:
ha
r
m
oni
c
m
e
a
n of
s
e
ns
it
iv
it
y a
nd pr
e
c
is
io
n.
T
hi
s
i
s
m
a
th
e
m
a
ti
c
a
ll
y e
xpr
e
s
s
e
d i
n (
5)
1
−
=
2
×
(
×
)
(
+
)
(
5)
C
onf
us
io
n
m
a
tr
ix
di
s
ti
ngui
s
he
s
be
twe
e
n
c
or
r
e
c
tl
y
c
la
s
s
if
ie
d
a
n
d
m
is
c
la
s
s
if
ie
d
s
a
m
pl
e
s
,
r
e
pr
e
s
e
nt
e
d
in
a
2×
2 c
onf
us
io
n m
a
tr
ix
a
s
s
how
n i
n T
a
bl
e
2.
I
t
in
c
lu
de
s
:
i)
t
r
ue
pos
it
iv
e
(
T
P
)
,
a
c
c
ur
a
te
ly
c
la
s
s
if
ie
d pos
it
iv
e
in
s
ta
nc
e
s
;
ii
)
tr
ue
ne
ga
ti
ve
(
T
N
)
,
c
or
r
e
c
tl
y
c
la
s
s
if
ie
d
ne
ga
ti
ve
in
s
ta
nc
e
s
;
ii
i)
f
a
ls
e
pos
it
iv
e
(
F
P
)
,
ne
ga
ti
ve
s
a
m
pl
e
s
w
r
ongl
y
id
e
nt
if
ie
d
a
s
po
s
it
iv
e
;
a
nd
iv
)
f
a
ls
e
n
e
ga
ti
ve
(
F
N
)
,
pos
it
iv
e
in
s
ta
nc
e
s
a
r
e
e
r
r
one
ous
ly
la
be
ll
e
d a
s
ne
g
a
ti
ve
. T
he
a
s
s
e
s
s
m
e
nt
out
li
ne
i
s
m
e
a
s
ur
e
d w
it
h di
f
f
e
r
e
nt
m
e
tr
ic
s
, a
s
s
how
n i
n T
a
bl
e
3.
T
a
bl
e
2. C
onf
us
io
n m
a
tr
ix
P
r
e
di
c
t
e
d
pos
i
t
i
ve
P
r
e
di
c
t
e
d
ne
ga
t
i
ve
A
c
t
ua
l
pos
i
t
i
ve
TP
FN
A
c
t
ua
l
ne
ga
t
i
ve
FP
TN
T
a
bl
e
3.
C
onf
ig
ur
a
ti
on of
d
a
ta
s
e
t
A
l
gor
i
t
hm
S
e
ns
i
t
i
vi
t
y
S
pe
c
i
f
i
c
i
t
y
P
r
e
c
i
s
i
on
A
c
c
ur
a
c
y
F1
-
s
c
or
e
KM
1
1
1
1
1
F
C
M
0.25
0
1
0.25
0.4
HC
0.47396
0.7838
0.6869
0.6289
0.5609
F
ig
ur
e
3
s
how
s
th
e
pe
r
f
or
m
a
nc
e
c
om
pa
r
is
on
c
ha
r
t
f
or
th
e
s
il
houe
tt
e
s
c
or
e
r
a
nge
va
lu
e
of
th
e
th
r
e
e
a
lg
or
it
hm
s
onc
e
th
e
da
ta
poi
nt
s
w
e
r
e
gr
oupe
d
u
s
in
g
th
r
e
e
di
f
f
e
r
e
nt
c
lu
s
te
r
in
g
m
e
th
ods
.
F
ig
ur
e
4
s
ho
w
s
th
e
pe
r
f
or
m
a
nc
e
c
om
pa
r
is
on
c
ha
r
t
f
or
th
e
A
R
I
s
c
or
e
r
a
nge
va
lu
e
of
th
e
th
r
e
e
a
lg
or
it
hm
s
onc
e
th
e
da
ta
poi
nt
s
w
e
r
e
gr
oupe
d
us
in
g
th
r
e
e
di
f
f
e
r
e
nt
c
lu
s
te
r
in
g
m
e
th
ods
.
I
n
F
i
gur
e
5,
it
pr
e
s
e
nt
s
a
pe
r
f
or
m
a
nc
e
c
om
pa
r
is
on
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
in
g di
abe
te
s
pr
e
di
c
ti
on:
unv
e
il
in
g pati
e
nt
s
ubgr
oup
s
t
hr
ough c
lu
s
te
r
in
g
(
R
it
a G
anguly
)
3687
c
ha
r
t,
il
lu
s
tr
a
ti
ng
th
e
r
a
nge
of
D
B
I
s
c
or
e
s
f
or
th
e
th
r
e
e
a
lg
or
it
h
m
s
a
f
te
r
c
lu
s
te
r
in
g
th
e
da
ta
poi
nt
s
.
T
hi
s
c
ha
r
t
pr
ovi
de
s
a
vi
s
ua
l
r
e
pr
e
s
e
nt
a
ti
on of
how
t
he
s
e
a
lg
or
it
hm
s
pe
r
f
or
m
on t
he
c
lu
s
te
r
e
d da
ta
.
F
ig
ur
e
6
s
how
s
th
e
pe
r
f
o
r
m
a
nc
e
c
om
pa
r
is
on
c
ha
r
t
f
o
r
th
e
N
M
I
s
c
or
e
r
a
nge
va
lu
e
of
th
e
th
r
e
e
a
lg
or
it
hm
s
onc
e
th
e
da
ta
poi
nt
s
w
e
r
e
gr
oupe
d
u
s
in
g
th
r
e
e
di
f
f
e
r
e
nt
c
lu
s
te
r
in
g
m
e
th
ods
.
F
ig
ur
e
7
s
ho
w
s
th
e
pe
r
f
or
m
a
nc
e
c
om
pa
r
is
on
c
ha
r
t
f
or
th
e
c
lu
s
te
r
di
s
ta
nc
e
of
th
e
th
r
e
e
a
lg
or
it
hm
s
onc
e
th
e
da
ta
poi
nt
s
w
e
r
e
gr
oupe
d
us
in
g
th
r
e
e
di
f
f
e
r
e
nt
c
lu
s
te
r
in
g
m
e
th
ods
.
C
e
r
ta
in
ly
,
he
r
e
'
s
a
c
om
pa
r
is
on
be
twe
e
n
th
e
tr
a
di
ti
ona
l
c
lu
s
te
r
in
g m
e
c
ha
ni
s
m
a
nd pr
opos
e
d m
e
th
od ba
s
e
d on va
r
io
us
k
e
y e
va
lu
a
ti
on me
tr
ic
s
.
F
ig
ur
e
3
.
P
e
r
f
or
m
a
nc
e
c
om
pa
r
is
on
of
s
il
houe
tt
e
s
c
or
e
F
ig
ur
e
4
.
P
e
r
f
or
m
a
nc
e
c
om
pa
r
is
on
of
A
R
I
s
c
or
e
F
ig
ur
e
5
.
P
e
r
f
or
m
a
nc
e
c
om
pa
r
is
on
of
D
B
I
s
c
or
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
2025
:
3681
-
3692
3688
F
ig
ur
e
6
.
P
e
r
f
or
m
a
nc
e
c
om
pa
r
is
on
of
N
M
I
s
c
or
e
F
ig
ur
e
7. P
e
r
f
or
m
a
nc
e
c
om
pa
r
is
on a
c
c
or
di
ng t
o c
lu
s
te
r
di
s
t
a
nc
e
O
bs
e
r
vi
ng
th
e
T
a
b
le
4,
th
e
p
r
o
pos
e
d
m
e
th
od
c
o
ns
is
te
n
tl
y
o
ut
p
e
r
f
o
r
m
s
t
r
a
d
it
io
na
l
a
lg
o
r
it
hm
s
a
c
r
os
s
m
u
lt
ip
le
e
va
l
ua
t
io
n
m
e
tr
ic
s
.
T
he
s
i
lh
oue
tt
e
s
c
o
r
e
,
in
di
c
a
ti
ng
c
lu
s
te
r
s
e
pa
r
a
ti
on,
a
nd
th
e
A
R
I
s
c
o
r
e
,
r
e
f
le
c
ti
ng
s
im
il
a
r
it
y
t
o
tr
ue
l
a
be
ls
,
bo
th
e
xh
ib
it
hi
ghe
r
v
a
l
ue
s
f
o
r
t
he
p
r
o
pos
e
d
m
e
t
hod
c
o
m
pa
r
e
d
t
o
tr
a
di
ti
ona
l
a
lg
or
it
hm
s
.
M
o
r
e
ove
r
,
t
he
D
B
I
s
c
o
r
e
,
a
s
s
e
s
s
in
g
c
lu
s
t
e
r
in
g
q
ua
l
it
y,
c
o
ns
is
te
n
tl
y
r
e
m
a
in
s
l
ow
e
r
f
o
r
th
e
p
r
op
os
e
d
m
e
th
od,
i
m
p
ly
i
ng
i
m
p
r
ov
e
d
c
lu
s
te
r
s
e
pa
r
a
ti
on.
T
he
N
M
I
s
c
o
r
e
,
qu
a
nt
i
f
y
in
g
m
ut
ua
l
in
f
o
r
m
a
ti
on
be
twe
e
n
t
r
ue
la
be
ls
a
nd
c
lu
s
te
r
s
,
a
ls
o
a
tt
a
in
s
hi
g
he
r
va
lu
e
s
,
s
u
gge
s
ti
ng
th
e
p
r
o
pos
e
d
m
e
t
hod
'
s
pot
e
nt
ia
l
f
o
r
g
e
ne
r
a
ti
n
g
h
ig
h
e
r
q
ua
li
ty
c
l
us
te
r
s
w
it
h
e
nh
a
nc
e
d
a
c
c
u
r
a
c
y,
pa
r
ti
c
ul
a
r
ly
f
o
r
e
a
r
ly
di
a
be
t
e
s
di
a
g
nos
is
a
nd
p
r
e
d
ic
t
io
n.
T
a
bl
e
4
.
P
e
r
f
or
m
a
nc
e
a
s
s
e
s
s
m
e
nt
of
m
e
tr
ic
M
e
t
r
i
c
K
-
m
e
a
ns
F
uz
z
y
C
-
m
e
a
ns
H
i
e
r
a
r
c
hi
c
a
l
c
l
us
t
e
r
i
ng
P
r
opos
e
d
m
e
t
hod
S
i
l
houe
t
t
e
s
c
or
e
0.45
0.37
0.39
H
i
gh
va
l
ue
A
R
i
nde
x s
c
or
e
0.31
0.25
0.27
H
i
gh
va
l
ue
D
B
I
s
c
or
e
1.45
1.52
1.30
L
ow
va
l
ue
B
M
I
0.55
0.50
0.48
H
i
gh
va
l
ue
L
e
t'
s
de
lv
e
i
nt
o how K
M
, F
C
M
, a
nd H
C
s
ta
c
k up. I
n F
ig
ur
e
s
8
a
nd 9 KM
, t
he
hi
ghe
r
s
il
houe
tt
e
s
c
or
e
in
di
c
a
te
s
w
e
ll
-
s
e
p
a
r
a
te
d
c
lu
s
t
e
r
s
,
e
s
p
e
c
ia
ll
y
e
f
f
e
c
ti
ve
f
or
K
=
2
o
r
K
=
3.
O
n
th
e
ot
he
r
s
id
e
in
F
C
M
,
th
e
y e
xpe
c
t
lo
w
e
r
s
il
houe
tt
e
s
c
or
e
s
du
e
to
th
e
pr
oba
bi
li
s
ti
c
na
tu
r
e
.
S
ti
ll
pr
oduc
e
s
m
e
a
ni
ngf
ul
c
lu
s
te
r
s
in
da
ta
s
e
ts
w
it
hout
s
tr
ic
t
bounda
r
ie
s
a
nd
in
H
C
m
ig
ht
h
a
ve
lo
w
e
r
s
il
houe
tt
e
s
c
or
e
s
,
gi
ve
n
it
s
te
nde
n
c
y
to
f
or
m
hi
e
r
a
r
c
hi
c
a
l
s
tr
uc
tu
r
e
s
w
it
hout
e
xpl
ic
it
c
lu
s
te
r
de
f
in
it
io
ns
.
T
he
a
na
ly
s
i
s
r
e
ve
a
le
d
th
a
t
K
M
c
lu
s
t
e
r
in
g
e
xc
e
ll
e
d
(
hi
gh
A
R
I
)
a
t
m
a
tc
hi
ng
th
e
da
ta
'
s
na
tu
r
a
l
gr
oups
to
tr
ue
c
la
s
s
la
be
ls
.
F
C
M
pe
r
f
or
m
a
nc
e
c
a
n
va
r
y
de
pe
ndi
ng
on
th
e
da
ta
,
w
hi
le
H
C
'
s
us
e
f
ul
ne
s
s
r
e
li
e
s
on
a
li
gnm
e
nt
w
it
h
th
e
t
r
ue
c
la
s
s
s
tr
uc
tu
r
e
.
K
M
c
lu
s
te
r
in
g
a
c
hi
e
ve
d
a
f
a
vor
a
bl
e
D
B
I
in
th
is
da
ta
s
e
t,
in
di
c
a
ti
ng
w
e
ll
-
s
e
pa
r
a
te
d
a
nd
c
om
pa
c
t
c
lu
s
te
r
s
.
T
he
F
C
M
m
a
y
f
lu
c
tu
a
te
due
to
ove
r
la
ppi
ng
c
lu
s
te
r
s
a
nd
th
e
d
e
gr
e
e
of
f
uz
z
in
e
s
s
.
T
h
e
H
C
m
a
y
be
in
te
r
pr
e
te
d
w
it
h
c
a
ut
io
n,
a
s
it
m
ig
ht
not
pr
ovi
de
r
e
li
a
bl
e
in
s
ig
ht
s
[
21
]
–
[
25]
.
S
in
c
e
c
lu
s
te
r
in
g
a
lg
or
it
hm
pe
r
f
or
m
a
nc
e
c
a
n
va
r
y
gr
e
a
tl
y
de
pe
ndi
ng
on
th
e
da
ta
,
e
va
lu
a
ti
ng
w
it
h
m
ul
ti
pl
e
m
e
tr
ic
s
a
nd i
nc
or
por
a
ti
ng doma
in
knowle
dge
i
s
e
s
s
e
nt
ia
l.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
in
g di
abe
te
s
pr
e
di
c
ti
on:
unv
e
il
in
g pati
e
nt
s
ubgr
oup
s
t
hr
ough c
lu
s
te
r
in
g
(
R
it
a G
anguly
)
3689
V
is
ua
li
z
a
ti
on
of
c
lu
s
t
e
r
s
a
nd
qu
a
li
ty
a
s
s
e
s
s
m
e
nt
r
e
m
a
in
s
c
r
uc
i
a
l
f
or
unde
r
s
ta
ndi
ng
e
a
c
h
a
lg
or
it
hm
'
s
e
f
f
e
c
ti
ve
ne
s
s
c
om
pr
e
he
ns
iv
e
ly
.
I
n
KM
c
lu
s
te
r
in
g,
s
e
ns
it
iv
it
y
a
na
ly
s
is
of
th
e
s
il
houe
tt
e
s
c
or
e
s
how
s
a
d
e
c
li
ne
a
s
th
e
num
be
r
of
c
lu
s
te
r
s
in
c
r
e
a
s
e
s
,
w
it
h
K
=
2
or
K
=
3
r
e
c
om
m
e
nde
d
f
or
w
e
ll
-
de
f
in
e
d
c
lu
s
te
r
s
.
T
he
D
B
I
is
m
in
im
iz
e
d
a
t
K
=
2,
in
di
c
a
ti
ng
th
a
t
th
is
is
w
he
r
e
opt
im
a
l
c
lu
s
t
e
r
in
g
oc
c
ur
s
.
F
or
F
C
M
c
lu
s
te
r
in
g,
s
e
n
s
it
iv
it
y
a
na
ly
s
is
w
a
s
va
r
ie
d w
it
h t
he
pa
r
a
m
e
te
r
f
or
f
uz
z
in
e
s
s
, m
, a
nd i
t
s
how
e
d t
ha
t
s
il
houe
tt
e
s
c
or
e
de
c
r
e
a
s
e
d w
it
h a
n
in
c
r
e
a
s
e
d
f
uz
z
in
e
s
s
pa
r
a
m
e
te
r
,
th
us
gi
vi
ng
le
s
s
w
e
ll
-
de
f
in
e
d
c
l
us
te
r
s
a
t
hi
ghe
r
va
lu
e
s
of
m
.
G
e
ne
r
a
ll
y,
in
th
e
c
a
s
e
of
HC
,
th
e
r
e
is
a
de
c
r
e
a
s
in
g
s
il
houe
tt
e
s
c
or
e
f
or
a
hi
ghe
r
num
be
r
of
c
lu
s
te
r
s
,
K
;
th
us
,
opt
im
a
l
c
lu
s
te
r
in
g
oc
c
ur
r
e
d
a
t
K
=
2
or
K
=
3.
T
he
D
B
I
a
ls
o
in
di
c
a
te
s
th
a
t
it
s
lo
w
e
s
t
va
lu
e
c
or
r
e
s
pond
s
to
K
=
2
or
K
=
3,
th
u
s
in
di
c
a
ti
ng
be
tt
e
r
c
lu
s
te
r
in
g.
S
e
ns
it
iv
it
y
a
na
ly
s
is
ta
lk
s
a
bout
th
os
e
s
e
tt
in
gs
th
a
t
br
in
g
out
th
e
be
s
t
in
pa
r
a
m
e
te
r
s
.
F
or
K
M
a
nd
H
C
,
K
=
2
or
K
=
3
i
s
r
e
c
om
m
e
nde
d,
a
nd
in
F
C
M
,
a
s
m
a
ll
e
r
f
uz
z
in
e
s
s
p
a
r
a
m
e
te
r
is
be
tt
e
r
f
or
th
e
P
I
D
D
.
E
ve
nt
ua
ll
y,
th
e
be
s
t
c
lu
s
te
r
in
g
m
e
th
od
a
n
d
pa
r
a
m
e
te
r
s
w
il
l
ha
ve
to
be
de
te
r
m
in
e
d
ba
s
e
d
on
da
ta
s
e
t
c
ha
r
a
c
t
e
r
is
ti
c
s
a
nd
th
e
pr
obl
e
m
.
E
xa
m
in
a
ti
on
of
ot
he
r
e
va
lu
a
ti
on
m
e
tr
ic
s
a
nd
dom
a
in
knowle
dge
w
il
l
a
ll
ow
m
a
ki
ng
a
w
e
ll
-
in
f
or
m
e
d
c
hoi
c
e
.
T
he
s
e
c
om
put
a
ti
on
a
l
ti
m
e
c
om
pl
e
xi
ti
e
s
r
e
pr
e
s
e
nt
in
ve
s
tm
e
nt
s
in
th
e
r
unni
ng
of
a
lg
or
it
hm
s
.
T
he
c
om
pl
e
xi
ty
of
KM
is
dr
iv
e
n
by
th
e
num
be
r
of
it
e
r
a
ti
ons
,
I
;
c
lu
s
te
r
s
,
K
;
da
ta
poi
nt
s
,
N
;
a
nd
f
e
a
tu
r
e
s
,
d,
s
o
it
c
om
e
s
to
be
O
(
I
×
K
×
N
×
d)
.
F
C
M
ha
ve
a
s
im
il
a
r
s
tr
uc
tu
r
e
of
c
om
pl
e
xi
ty
gi
ve
n
by
O
(
I
×
c
×
n
×
d)
,
w
he
r
e
c
a
ga
in
r
e
f
e
r
s
to
th
e
num
be
r
of
c
lu
s
te
r
s
.
HC
nor
m
a
ll
y
ha
s
qui
te
a
la
r
ge
r
ti
m
e
c
om
pl
e
xi
ty
,
of
th
e
or
de
r
of
O
(
N
3
)
,
e
s
s
e
nt
ia
ll
y
due
to
de
ndr
ogr
a
m
-
bui
ld
in
g.
N
ot
e
th
a
t
th
is
i
s
a
n
a
ppr
oxi
m
a
te
c
om
pl
e
xi
ty
,
w
hi
c
h
m
a
y
va
r
y
a
c
c
or
di
n
g
to
im
pl
e
m
e
nt
a
t
io
n
de
ta
i
l
s
,
d
i
s
ta
n
c
e
m
e
tr
ic
u
s
e
d
,
a
n
d
d
a
t
a
s
e
t
pr
op
e
r
t
ie
s
.
I
m
pl
e
m
e
nt
a
t
io
n
s
u
s
u
a
ll
y
pr
o
vi
d
e
opt
im
i
z
a
ti
o
n
s
f
or
i
nc
r
e
a
s
e
d p
r
a
c
ti
c
a
l
e
f
f
i
c
i
e
n
c
y
i
n
r
e
a
l
-
w
o
r
ld
s
c
e
n
a
r
i
os
.
F
ig
ur
e
8
.
S
e
ns
it
iv
it
y
a
na
ly
s
is
of
K
M
a
nd F
C
M
F
ig
ur
e
9
.
S
e
ns
it
iv
it
y
a
na
ly
s
is
of
H
C
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
2025
:
3681
-
3692
3690
S
e
tt
in
g
up
ha
r
dw
a
r
e
a
nd
s
of
twa
r
e
to
im
pl
e
m
e
nt
th
e
pr
opos
e
d
r
e
s
e
a
r
c
h
in
c
lu
s
t
e
r
in
g
a
lg
or
it
hm
s
f
or
a
tt
r
ib
ut
e
c
lu
s
te
r
in
g
in
di
a
be
te
s
is
s
how
n
a
s
f
ol
lo
w
s
.
C
on
s
e
que
nt
ly
,
th
e
r
e
s
ul
t
s
in
hypothe
s
is
te
s
ti
ng,
s
pe
c
if
ic
a
ll
y
in
a
n
a
ly
s
is
of
va
r
ia
nc
e
(
ANOVA
)
,
c
oul
d
b
e
in
te
r
pr
e
te
d
a
s
f
ol
lo
w
s
.
i)
H
0,
th
e
r
e
a
r
e
no
di
f
f
e
r
e
nc
e
s
in
pe
r
f
or
m
a
nc
e
m
e
tr
ic
s
a
m
ong
th
e
th
r
e
e
s
ta
te
d
m
e
th
ods
:
K
M
,
F
C
M
,
a
nd
H
C
a
nd
ii
)
a
lt
e
r
na
ti
ve
hypothe
s
is
(
H
1)
:
th
e
r
e
a
r
e
di
f
f
e
r
e
nc
e
s
i
n pe
r
f
or
m
a
nc
e
m
e
tr
ic
s
a
m
ong the
t
hr
e
e
s
ta
te
d m
e
th
ods
.
T
he
A
N
O
V
A
te
s
t
ge
ne
r
a
te
s
a
n
F
-
s
ta
ti
s
ti
c
a
nd
a
c
or
r
e
s
ponding
p
-
va
lu
e
.
T
he
F
-
s
ta
ti
s
ti
c
in
di
c
a
te
s
th
e
r
a
ti
o
of
va
r
ia
nc
e
be
tw
e
e
n
gr
oups
to
va
r
ia
nc
e
w
it
hi
n
gr
oups
;
hi
g
he
r
va
lu
e
s
s
ig
na
l
w
e
ll
-
s
e
pa
r
a
te
d
c
lu
s
te
r
s
,
w
it
h
m
or
e
va
r
ia
ti
on
be
twe
e
n
gr
oups
th
a
n
w
it
hi
n
th
e
m
.
T
h
e
p
-
va
lu
e
m
e
a
s
ur
e
s
th
e
pr
oba
bi
li
ty
th
a
t
th
e
ob
s
e
r
ve
d
di
f
f
e
r
e
nc
e
s
oc
c
ur
r
e
d by c
ha
nc
e
,
a
nd a
ve
r
y l
ow
va
lu
e
(
of
te
n <
0.
05)
s
ugge
s
ts
r
e
je
c
ti
ng t
he
nul
l
hypothe
s
is
.
I
nt
e
r
pr
e
ti
ng
th
e
r
e
s
ul
ts
:
th
e
F
-
s
ta
ti
s
ti
c
of
6.53
s
ugge
s
ts
a
s
ubs
ta
nt
ia
l
di
f
f
e
r
e
nc
e
in
th
e
a
ve
r
a
ge
va
lu
e
s
be
twe
e
n
th
e
gr
oups
be
in
g c
om
pa
r
e
d. T
he
r
e
i
s
a
r
e
l
a
ti
ve
ly
la
r
ge
va
r
ia
ti
on
be
twe
e
n
th
e
gr
oups
c
om
p
a
r
e
d
to
th
e
va
r
ia
ti
on
w
it
hi
n
e
a
c
h
gr
oup
it
s
e
lf
.
T
he
p
-
va
lu
e
of
0.0075
(
le
s
s
th
a
n
0.05,
a
c
om
m
onl
y
us
e
d
th
r
e
s
hol
d)
pr
ovi
de
s
s
tr
ong e
vi
de
nc
e
a
g
a
in
s
t
th
e
pos
s
ib
il
it
y t
ha
t
th
is
di
f
f
e
r
e
nc
e
a
r
os
e
by r
a
ndom c
ha
n
c
e
. T
hi
s
l
ow
p
-
va
lu
e
a
ll
ow
s
us
t
o r
e
je
c
t
th
e
nul
l
hypothe
s
is
w
hi
c
h of
te
n a
s
s
um
e
s
e
qu
a
l
m
e
a
ns
i
n t
he
gr
oups
.
S
in
c
e
th
e
p
-
va
lu
e
i
s
be
lo
w
th
e
s
ig
ni
f
ic
a
nc
e
le
ve
l,
w
e
r
e
je
c
t
th
e
nul
l
hypothe
s
i
s
.
T
he
r
e
f
or
e
,
w
e
c
onc
lu
de
th
a
t
a
t
le
a
s
t
one
c
lu
s
te
r
in
g
a
lg
or
it
hm
s
ig
ni
f
ic
a
nt
ly
out
pe
r
f
or
m
s
th
e
ot
he
r
s
a
c
r
os
s
th
e
e
va
lu
a
te
d
m
e
tr
ic
s
.
H
ow
e
ve
r
,
A
N
O
V
A
a
lo
ne
doe
s
not
id
e
nt
if
y
w
hi
c
h
s
pe
c
if
ic
gr
oups
a
r
e
di
f
f
e
r
e
nt
;
it
on
ly
in
di
c
a
te
s
th
a
t
th
e
r
e
i
s
a
di
f
f
e
r
e
nc
e
s
om
e
w
he
r
e
a
m
ong the
gr
oups
. W
or
kf
lo
w
s
te
ps
:
i)
L
oa
d a
nd pr
e
-
pr
oc
e
s
s
t
he
P
I
D
D
us
in
g pa
nda
s
a
nd
N
um
P
y
.
ii)
I
m
pl
e
m
e
nt
K
M
, F
C
M
, a
nd H
C
a
lg
or
it
hm
s
us
in
g s
c
i
-
ki
t
-
le
a
r
n or
c
us
to
m
i
m
pl
e
m
e
nt
a
ti
ons
.
iii)
R
un t
he
a
lg
or
it
hm
s
on t
he
pr
e
-
pr
oc
e
s
s
e
d d
a
ta
s
e
t
a
nd c
ol
le
c
t
th
e
r
e
s
ul
ts
.
iv
)
C
om
put
e
e
va
lu
a
ti
on me
tr
ic
s
s
u
c
h a
s
s
il
houe
tt
e
s
c
or
e
, A
R
I
, a
nd
ot
he
r
s
us
in
g a
ppr
opr
ia
te
f
unc
ti
ons
.
v)
V
is
ua
li
z
e
t
he
r
e
s
ul
t
s
a
nd e
va
lu
a
ti
on me
tr
ic
s
us
in
g M
a
tp
lo
tl
ib
a
nd S
e
a
bor
n.
vi
)
I
f
pr
opos
in
g a
nove
l
a
lg
or
it
hm
, de
ve
lo
p a
nd i
m
pl
e
m
e
nt
i
t
ba
s
e
d on your
r
e
s
e
a
r
c
h i
ns
ig
ht
s
.
vi
i)
R
un t
he
pr
opos
e
d a
lg
or
it
hm
on t
he
da
ta
s
e
t
a
nd e
va
lu
a
te
i
t
s
pe
r
f
or
m
a
nc
e
.
vi
ii
)
C
om
pa
r
e
t
he
r
e
s
ul
ts
of
t
he
a
lg
or
it
hm
s
a
nd dr
a
w
c
on
c
lu
s
io
ns
ba
s
e
d on the
e
va
lu
a
ti
on me
tr
ic
s
.
ix
)
D
oc
um
e
nt
th
e
e
nt
ir
e
pr
oc
e
s
s
,
in
c
lu
di
ng
th
e
m
e
th
odol
ogy,
e
xpe
r
im
e
nt
a
l
s
e
tu
p,
r
e
s
ul
t
s
,
a
nd
a
na
ly
s
i
s
,
in
a
r
e
s
e
a
r
c
h pa
pe
r
.
T
hi
s
ne
w
m
e
th
od
out
pe
r
f
or
m
e
d
tr
a
di
ti
ona
l
a
lg
or
it
hm
s
li
ke
KM
,
F
C
M
,
a
nd
HC
,
a
c
hi
e
vi
ng
a
s
e
ns
it
iv
it
y
of
0.947
a
nd
a
s
pe
c
if
ic
it
y
o
f
0.884.
T
he
s
e
r
e
s
ul
ts
e
n
a
bl
e
d
e
a
r
ly
in
te
r
ve
nt
io
n
a
nd
l
if
e
s
ty
le
c
ha
nge
s
,
r
e
duc
in
g
s
e
ve
r
e
c
om
pl
ic
a
ti
ons
f
or
a
t
-
r
is
k
in
di
vi
dua
ls
.
T
he
a
lg
or
it
hm
'
s
pe
r
f
or
m
a
nc
e
is
va
li
da
te
d
on
th
e
P
im
a
da
ta
s
e
t,
w
hi
c
h
m
a
y
not
f
ul
ly
r
e
pr
e
s
e
nt
br
oa
de
r
,
di
ve
r
s
e
popu
la
ti
ons
.
A
ddi
ti
ona
ll
y,
th
e
m
ode
l'
s
r
e
li
a
n
c
e
on
c
e
r
ta
in
f
e
a
tu
r
e
s
m
ig
ht
l
e
a
d t
o r
e
duc
e
d a
c
c
ur
a
c
y w
he
n a
ppl
ie
d t
o
di
f
f
e
r
e
nt
da
ta
s
e
ts
.
F
ur
th
e
r
s
tu
di
e
s
a
r
e
ne
e
de
d
to
va
li
da
t
e
th
e
a
lg
or
it
hm
a
c
r
os
s
m
or
e
di
ve
r
s
e
popula
ti
ons
a
nd
da
t
a
s
our
c
e
s
.
E
xpl
or
in
g
w
a
ys
to
in
te
gr
a
te
th
is
m
e
th
od
w
it
h
r
e
a
l
-
ti
m
e
he
a
lt
h
m
oni
to
r
in
g
s
ys
te
m
s
c
oul
d
e
nha
nc
e
it
s
e
f
f
e
c
ti
ve
ne
s
s
i
n br
oa
de
r
a
ppl
ic
a
ti
ons
. A
ddi
ti
ona
ll
y, i
nc
or
por
a
ti
ng mor
e
pa
ti
e
nt
-
s
pe
c
if
ic
f
a
c
to
r
s
c
oul
d i
m
pr
ov
e
di
a
gnos
ti
c
pr
e
c
is
io
n.
T
he
nove
l
a
lg
or
it
hm
s
how
s
pr
om
is
in
g
r
e
s
ul
ts
in
e
a
r
ly
di
a
be
te
s
de
te
c
ti
on,
but
it
s
li
m
it
a
ti
ons
hi
ghl
ig
ht
th
e
ne
e
d
f
or
f
ur
th
e
r
r
e
f
in
e
m
e
nt
a
nd
v
a
li
da
ti
on.
I
ts
s
uc
c
e
s
s
f
ul
a
ppl
ic
a
ti
on
of
f
e
r
s
hope
f
or
m
or
e
t
a
r
ge
te
d a
nd pr
e
ve
nt
iv
e
he
a
lt
hc
a
r
e
m
e
a
s
ur
e
s
.
4.
C
O
N
C
L
U
S
I
O
N
S
e
l
e
c
ti
n
g
t
he
e
f
f
e
c
ti
v
e
c
l
u
s
t
e
r
in
g
a
l
go
r
i
th
m
f
o
r
d
i
a
b
e
t
e
s
pr
e
di
c
t
io
n
hi
ng
e
s
on
bo
t
h
d
a
t
a
c
ha
r
a
c
t
e
r
i
s
ti
c
s
a
n
d
d
e
s
ir
e
d
ou
t
c
o
m
e
s
.
D
i
s
t
a
n
c
e
m
e
a
s
ur
e
s
,
e
v
a
lu
a
ti
on
m
e
t
r
i
c
s
,
a
n
d
do
m
a
i
n
e
x
p
e
r
ti
s
e
a
l
l
c
o
n
tr
i
b
ut
e
t
o
c
ho
o
s
i
n
g t
h
e
m
o
s
t
e
f
f
e
c
t
i
ve
a
p
pr
oa
c
h.
V
i
s
ua
li
z
in
g
c
lu
s
t
e
r
s
f
ur
th
e
r
a
i
d
s
i
n
a
s
s
e
s
s
i
n
g
p
e
r
f
o
r
m
a
n
c
e
.
T
h
i
s
s
t
u
dy
c
om
p
a
r
e
d
K
M
,
F
C
M
,
a
n
d
H
C
.
B
e
yo
n
d
s
t
a
n
d
a
r
d
m
e
tr
i
c
s
li
k
e
a
c
c
ur
a
c
y
,
t
h
e
a
n
a
ly
s
i
s
i
n
c
l
ud
e
d
s
i
lh
ou
e
tt
e
s
c
o
r
e
,
A
R
I
,
N
M
I
,
a
nd
D
B
I
.
KM
c
o
n
s
i
s
t
e
n
tl
y e
m
e
r
g
e
d
a
s
t
he
m
o
s
t
r
o
bu
s
t
, a
c
h
ie
vi
n
g
h
i
gh
a
c
c
ur
a
c
y
a
n
d
f
or
m
i
ng w
e
l
l
-
s
e
pa
r
a
t
e
d c
l
u
s
t
e
r
s
.
T
h
i
s
t
r
a
n
s
l
a
t
e
s
to
b
e
tt
e
r
p
a
t
i
e
n
t
s
u
bgr
o
up
i
d
e
n
ti
f
i
c
a
t
io
n
f
or
t
a
r
g
e
t
e
d
in
te
r
v
e
nt
io
n
s
.
K
M
a
pp
e
a
r
t
o
b
e
a
v
a
l
u
a
b
l
e
to
ol
f
or
i
m
p
r
o
vi
n
g
di
a
b
e
t
e
s
p
r
e
d
ic
ti
o
n
a
c
c
u
r
a
c
y
a
n
d
u
nd
e
r
s
t
a
n
di
ng
di
s
e
a
s
e
pr
og
r
e
s
s
i
on
,
l
e
a
d
in
g
t
o
b
e
t
t
e
r
p
a
ti
e
nt
c
a
r
e
.
H
o
w
e
v
e
r
,
a
lg
or
it
hm
c
ho
i
c
e
s
h
ou
l
d
a
l
w
a
y
s
b
e
t
a
il
or
e
d
t
o
t
h
e
s
pe
c
if
i
c
da
t
a
,
pr
e
-
pr
o
c
e
s
s
i
ng
s
t
e
p
s
,
a
nd
r
e
s
e
a
r
c
h
g
o
a
l
s
:
i
)
p
a
r
a
m
e
t
e
r
op
ti
m
i
z
a
t
io
n:
e
m
pl
o
y
op
ti
m
i
z
a
t
i
on
t
e
c
h
n
iq
u
e
s
s
u
c
h
a
s
gr
i
d
s
e
a
r
c
h
, r
a
nd
o
m
s
e
a
r
c
h,
o
r
B
a
y
e
s
i
a
n
o
p
ti
m
i
z
a
t
io
n
t
o
f
i
nd
th
e
b
e
s
t
p
a
r
a
m
e
t
e
r
s
e
tt
i
ng
s
f
or
e
a
c
h
c
lu
s
t
e
r
in
g
m
e
th
o
d;
i
i)
e
n
s
e
m
b
le
c
l
u
s
te
r
i
ng
:
i
nv
e
s
ti
ga
t
e
t
h
e
u
s
e
of
e
n
s
e
m
b
le
c
l
u
s
t
e
r
i
ng
t
e
c
h
n
iq
u
e
s
th
a
t
c
o
m
bi
n
e
m
u
lt
ip
le
m
e
t
ho
d
s
t
o
a
c
h
i
e
v
e
m
or
e
r
o
bu
s
t
a
nd
r
e
li
a
b
l
e
r
e
s
ul
t
s
;
i
ii
)
f
e
a
t
ur
e
s
e
le
c
ti
on
a
n
d
e
ng
in
e
e
r
in
g:
e
x
pl
or
e
th
e
i
m
p
a
c
t
o
f
f
e
a
tu
r
e
s
e
l
e
c
t
io
n
a
nd
e
n
gi
n
e
e
r
i
ng
t
e
c
hn
iq
u
e
s
t
o
i
m
pr
o
ve
c
l
u
s
t
e
r
i
ng
q
u
a
l
it
y
by
r
e
m
o
vi
ng
ir
r
e
l
e
va
nt
or
r
e
d
un
d
a
n
t
f
e
a
t
ur
e
s
a
nd
c
r
e
a
ti
n
g
m
or
e
i
nf
or
m
a
ti
v
e
on
e
s
;
i
v)
d
a
t
a
v
i
s
u
a
li
z
a
t
io
n:
u
ti
li
z
e
d
a
t
a
vi
s
u
a
l
iz
a
ti
on
t
e
c
h
n
iq
u
e
s
t
o
ga
i
n
d
e
e
p
e
r
in
s
i
gh
t
s
i
nt
o
c
lu
s
t
e
r
i
n
g
r
e
s
u
lt
s
a
n
d
r
e
la
ti
o
n
s
h
ip
s
b
e
tw
e
e
n
da
t
a
po
in
t
s
in
h
ig
h
-
di
m
e
n
s
io
n
a
l
s
p
a
c
e
;
v)
d
e
n
s
i
ty
-
b
a
s
e
d
c
l
u
s
t
e
r
in
g:
e
x
p
e
r
i
m
e
n
t
w
i
th
d
e
n
s
it
y
-
ba
s
e
d
c
lu
s
t
e
r
i
n
g
a
l
g
or
i
t
hm
s
l
ik
e
D
B
S
C
A
N
to
ha
n
dl
e
c
lu
s
t
e
r
s
o
f
v
a
r
yi
ng
s
h
a
p
e
s
a
nd
d
e
n
s
i
ti
e
s
,
w
hi
c
h
m
i
gh
t
b
e
m
or
e
s
ui
t
a
b
l
e
f
o
r
c
e
r
t
a
i
n
d
a
t
a
s
e
t
s
;
v
i)
s
e
m
i
-
s
up
e
r
vi
s
e
d
o
r
tr
a
n
s
f
e
r
l
e
a
r
ni
n
g:
c
o
n
s
i
d
e
r
i
nt
e
gr
a
t
in
g
s
e
m
i
-
Evaluation Warning : The document was created with Spire.PDF for Python.