Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
10
,
No.
5
,
Octo
be
r
2020
,
pp.
4853
~
4862
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
5
.
pp
4853
-
48
62
4853
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Fra
m
ework fo
r
e
ffici
ent
t
ra
nsf
or
mation f
or
c
omplex medi
cal
d
ata for
i
mp
ro
vin
g
a
n
alytical
c
ap
ability
Chandr
u
A
.
S
.
1
,
See
thara
m
K
.
2
1
Depa
rtment of
Com
pute
r
Scie
n
ce
and
Engi
ne
ering
,
Visvesvar
a
ya
T
ec
hnolog
ical
Univer
sit
y
(VT
U)
,
Indi
a
2
Depa
rtment of
Com
pute
r
Scie
n
ce
and
Engi
ne
ering
,
Jnana
v
ika
s
I
nstit
ute of Te
chn
olog
y
,
Ind
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
ug
14
, 201
9
Re
vised
Ma
r
3
,
2020
Accepte
d
Ma
r
1
8
, 202
0
The
adop
ti
on
of
var
ious t
e
chnol
o
gic
a
l
adv
anc
eme
nt
has
be
en
a
lrea
d
y
adopt
ed
in
the
ar
ea
of
hea
lt
h
ca
r
e
sec
tor.
Thi
s
adop
t
ion
fac
i
li
t
ates
invol
untar
y
gene
ra
ti
on
of
m
edi
ca
l
d
at
a
t
hat
c
an
be
au
t
onom
ously
pro
gra
m
m
ed
to
be
forwarde
d
t
o
a
dest
ine
d
h
ub
in
th
e
for
m
of
cl
oud
st
ora
ge
uni
ts.
How
eve
r,
owing
to
such
t
ec
hnol
ogie
s
th
ere
is
m
assive
form
at
ion
of
compl
e
x
m
edi
ca
l
d
at
a
th
at
significan
tly
ac
ts
as
an
ov
erh
ea
d
towar
ds
per
form
ing
ana
l
y
t
ical
oper
a
ti
on
as
well
as
unwante
d
storage
uti
l
izati
on
.
The
ref
ore
,
the
propose
d
s
y
stem
implements
a
novel
tra
nsf
orm
at
ion
te
chn
i
que
tha
t
is
ca
pab
le
of
usin
g
a
t
emplat
e
b
ase
d
stuct
ur
e
o
ver
cl
oud
for
gene
ra
ti
ng
struct
ure
d
data
f
rom
highly
unstr
uct
ure
d
data
in
a
non
-
conve
n
ti
on
al
m
anne
r
.
The
cont
ribu
ti
o
n
of
th
e
props
oed
m
et
hodolo
g
y
is
th
at
it
o
ffe
rs
faster
proc
essing
and
s
tora
ge
op
ti
m
izat
ion.
Th
e
stud
y
o
utc
om
e
al
so
pro
ves
thi
s
fa
ct
to
show
prop
soed
sche
m
e
exc
el
s
bet
te
r
in
per
for
m
anc
e
in
cont
ra
st
to
exi
sting
dat
a
tra
nsform
ati
on
sche
m
e.
Ke
yw
or
d
s
:
Data tra
ns
f
or
m
at
ion
Healt
hcar
e
a
na
ly
ti
cs
Me
dical
Stru
ct
ur
iz
at
io
n
Text m
ining
Copyright
©
202
0
Instit
ute of
Ad
v
ance
d
Engi
ne
eri
ng
and
Sc
ie
n
ce
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Chan
dru
A
.
S
.
,
Re
cearch
Sch
ol
ar,
Dep
a
rtm
ent Com
pu
te
r
Sci
ence a
nd
E
ng
i
neer
i
ng,
Visv
es
va
raya
Tech
no
l
og
ic
al
Un
i
ver
sit
y (
VT
U)
,
Be
la
gav
i
, In
dia
.
Em
a
il
: chan
dru
.as.r
e
searc
h@g
m
ai
l.co
m
1.
INTROD
U
CTION
W
it
h
the
incr
easi
ng
ad
opti
on
of
the
m
ob
il
e
netwo
r
k
an
d
ubiq
uitou
s
com
pu
ti
ng
,
th
ere
are
al
s
o
trem
end
ou
s
in
crease
in
a
pp
li
cat
ion
as
w
el
l
as
serv
ic
e
s
that
facil
it
at
e
the
us
e
r
with
exte
ns
iv
e
data
avail
abili
ty
[1
,
2].
Cl
oud
c
om
pu
ti
ng
m
a
kes
t
his
proc
ess
easi
er
by
their
dif
fer
e
nt
ty
pes
of
s
ta
nd
a
rd
serv
ic
es
[3
]
.
T
he
be
nef
ic
ia
l
f
act
or
of
this
te
chnolo
gies
is
that
there
is
a
great
er
deal
of
s
ynch
ronici
ty
a
m
on
g
diff
e
re
nt
infor
m
at
ion
captu
ri
ng
dev
ic
es
,
protoc
ols
us
e
d
f
or
pr
ocessin
g
the
inf
or
m
at
ion,
and
netw
ork
dev
ic
es
to
route
the
da
ta
to
def
ine
d
serv
e
rs
[4
,
5]
.
Howe
ver
,
th
ere
are
al
so
e
xtrem
el
y
challen
gi
ng
sit
uatio
n
for
handlin
g
s
uc
h
m
assive
fo
rm
s
of
e
vo
l
ving
data.
T
he
pro
po
s
ed
stu
dy
c
on
si
ders
a
cas
e
stud
y
of
he
a
lt
hcar
e
sect
or
w
hich
auto
no
m
ou
sly
gathe
r
data
from
diff
eren
t
f
or
m
s
of
analo
g,
di
gital
,
and
hybr
id
dev
ic
es
and
fetches
t
his
dat
a
to
the
hub
w
her
e
it
is
f
ur
t
he
r
sto
red
i
n
a
physi
cal
un
it
or
forw
a
r
ded
t
o
da
ta
center.
A
t
ypic
al
sm
art
he
al
thcare
unit
can
be
consi
de
red
t
o
handle
m
ulti
ple
fo
rm
s
of
da
ta
us
in
g
di
ff
e
r
ent
ty
pes
of
s
ensin
g
dev
ic
es
where
div
e
rsified
in
form
ation
a
bout t
he p
at
ie
nt
healt
h
is c
ollec
te
d [
6].
Howe
ver,
the
chall
en
ges
i
n
this
sm
art
te
chnolo
gical
a
dvancem
ent
are
i)
eac
h
s
e
nsor
capt
ur
es
a
con
ti
nu
ou
s
da
ta
and
the
siz
e
of
the
data
is
ever
inc
reasin
g
w
hich
ca
nnot
be
store
d
in
ph
ysi
cal
stora
ge
ov
e
r
the
pr
em
ise
s
of
healt
hca
re
un
it
s,
ii
)
the
fo
r
m
s
and
form
ats
of
the
data
a
re
highly
div
e
r
sifie
d
an
d
ther
e
cou
ld
be
m
or
e
inclusion
of
the
su
c
h
data
fr
om
the
sam
e
healt
hcare
un
it
s
le
adin
g
to
fo
rm
at
ion
of
com
plex
m
e
dical
data
[7
]
,
ii
i)
these
data
wh
e
n
thought
of
aggreg
at
e
d
in
one
place
is
qu
it
e
diff
ic
ult
to
be
store
d
as
they
form
a
conditi
on
of
highly
unstruct
ur
e
d
data
(in
s
uch
f
or
m
,
conven
ti
onal
kn
owle
dge
ext
racti
on
m
et
ho
d
ca
nnot
be
app
li
ed
),
iv)
e
xisti
ng
ap
proa
ches
has
la
rg
e
r
de
gr
ee
of
us
a
ge
of
distri
bu
t
ed
s
of
t
war
e
f
r
a
m
ewo
r
k
li
ke
Hado
op
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
53
-
48
62
4854
and
Ma
pRed
uc
e;
ho
we
ve
r,
they
are
ver
y
m
uch
baseli
ne
sta
nd
ar
d
an
d
there
is
al
ways
a
need
to
a
m
end
the
c
om
plex
arch
it
ect
ure
of
it
wh
e
n
t
he
a
pp
li
cat
io
n
de
m
and
s
c
ha
ng
e
s.
Be
side
s
the
re
a
re
m
any
r
eporte
d
loopholes
of
s
uch
f
ram
ewo
r
k
w
hic
h
can
not
be
rec
ommende
d
t
o
be
use
d
for
a
c
riti
cal
app
li
cat
ion
e.g.
healt
hcar
e
whi
ch
in
volves
se
ns
it
ive cli
nical
data to
be use
d i
n fu
t
ur
ist
ic
ana
ly
sis [8
]
.
Existi
ng
m
ining
a
ppr
oach
e
s
are
a
pp
li
ed
on
com
plex
data
but
with
high
ly
def
ine
d
a
nd
co
ntr
olled
env
i
ronm
ent
a
nd
thei
r
cost
e
ff
ect
ive
ness
in
the
fiel
d
of
m
edical
sci
en
ce
is
no
t
ye
t
been
pro
ve
n
[
9]
.
It
is
essenti
al
to
ens
ur
e
that
a
com
plex
f
or
m
of
m
edical
data
be
identifie
d
of
it
s
un
iq
ue
data
st
ru
ct
ur
e
s
o
that
there
is
a
good
possi
bili
ty
to
per
f
orm
str
uctur
iz
at
ion
of
the
c
omplex
m
edical
data.
On
ce
t
he
c
om
plex
m
edical
data
is
form
ulate
d
i
n
ef
fici
ent
dat
a
structu
re
tha
n
it
op
e
ns
up
m
any
ways
to
perform
analy
s
is
ov
e
r
the
str
uc
ture
d
data.
H
ow
e
ve
r,
it
is
no
t
eas
y
way
as
s
uch
data
are
al
wa
ys
in
the
f
or
m
of
s
tream
and
the
re
a
re
va
rio
us
associat
ed
pro
blem
s
[1
0].
A
no
t
her
chall
en
ging
sce
nar
i
o
of
c
om
plex
m
e
dical
data
will
be
to
pe
rfo
rm
stora
ge
op
ti
m
iz
ation
to
o
over clo
ud. Not all
the d
at
a sh
ould
be
sto
red
ov
e
r
a cloud which also
e
xponentia
ll
y i
ncr
eases
in
siz
e.
He
nc
e,
a
bette
r
stora
ge
optim
iz
at
ion
can
be
done
if
on
ly
the
m
ined
data
are
sto
re
d
in
the
distrib
uted
cl
oud
with
a
superi
or
f
or
m
of
ind
e
xing
m
echan
ism
.
Ther
ef
ore,
the
pr
opos
e
d
syst
e
m
pr
ese
nts
a
so
luti
on w
he
re
a
un
i
que
tra
ns
f
or
m
at
ion
syst
em
is
i
m
ple
m
ented
as
a
s
olu
ti
on
t
ow
a
r
d
cu
r
ving
unstr
uct
ured
data
to
t
he
highly
structu
re
d data wit
hout
us
i
ng
any of th
e syst
e
m
that has be
en
im
ple
m
ente
d
ti
ll
d
at
e or
usi
ng
syst
em
w
hi
ch ha
s
higher
res
ourc
e
de
pe
nd
e
ncie
s.
T
he
fo
c
us
is
to
wards
i
ncorpor
at
in
g
cost
eff
ect
ive
ness
towa
r
ds
norm
alizi
ng
the co
m
plex
m
edical
d
at
a.
The
pap
e
r
disc
us
ses
this
s
olu
t
ion
in
te
rm
s
of
m
et
ho
dolo
gy,
al
gorithm
,
and
be
nc
hm
ark
ed
outc
om
es.
The
orga
nizat
ion
of
this
p
a
pe
r
is
as
fo
ll
ows:
Sect
ion
1
disc
us
ses
a
bout
t
he
existi
ng
li
te
ratur
es
w
her
e
di
fferent
te
c
hn
iq
ues
a
re
discusse
d
for
detect
ion
sc
he
m
es
us
ed
i
n
powe
r
tra
ns
m
iss
ion
li
nes
f
ollo
wed
by
disc
ussi
on
of
researc
h
pro
bl
e
m
s
and
pro
posed.
Sect
io
n
2
discusse
s
ab
out
al
go
rithm
i
mp
le
m
entat
ion
fo
ll
ow
e
d
by
dis
cussion
of r
es
ult analy
s
is i
n
Sect
io
n 3. Fi
nally
, th
e
c
oncl
us
i
ve
rem
ark
s a
re
pro
vid
e
d
in
Secti
on
4.
The
Ba
ck
gro
und
.
This
sect
i
on
discu
sses
about
the
e
xisti
ng
data
tra
nsfo
rm
at
ion
an
d
proce
ssin
g
schem
es
carried
out
towa
rds
healt
hcar
e
sect
or.
Stu
dy
cons
iderin
g
the
E
HR
-
base
d
data
was
carrie
d
out
by
Muslim
et
al
.
[
11
]
fo
c
us
i
ng
towa
r
ds
redu
ci
ng
the
discr
et
ion
i
n
the
e
xisti
ng
sta
ndar
ds
of
m
edical
dat
a
.
Em
ph
asi
s
towa
rd
s
a
doptio
n
of
E
TL
(E
xtra
ct
/Transfo
rm
/L
oad)
to
wards
proces
sin
g
the
heter
og
e
ne
ou
s
data
is
al
so
discusse
d
by
Diouf
et
al
.
[
12
]
w
her
e
the
a
utho
rs
c
la
i
m
ed
tha
t
ex
ist
ing
tra
nsfo
r
m
at
ion
schem
e
sti
ll
encou
nters big
ger
c
halle
nges
in fro
m
o
f pr
e
s
ent stat
e of com
plex
data.
Wan
g et
al. [1
3] h
ave
d
e
velo
pe
d
a
dat
a
sh
ari
ng
schem
e
us
in
g
c
onve
ntion
al
E
TL
proces
s
over
va
ried
dataset
to
sh
ow
th
at
the
ir
w
ork
is
cap
able
of
ini
ti
at
ing
bette
r
tra
ns
f
or
m
at
ion
process.
Dif
fer
e
nt
w
ork
ha
s
bee
n
ca
rr
ie
d
ou
t
t
ow
a
r
ds
m
akin
g
the
healt
hcar
e
syst
e
m
s
m
art
fo
r
pr
eci
se
diagnosi
ng
[14
-
17]
.
The
wo
r
k
of
Ma
g
ari
no
et
al
.
[18]
has
presente
d
a
prot
otype
-
ba
sed
m
od
el
f
or
analy
zi
ng
t
he
act
ion
of
sle
ep
sta
te
.
Existi
ng
syst
e
m
has
al
so
bein
g
ca
rr
ie
d
ou
t
towa
rd
s le
ver
a
ging the dia
gn
os
is sy
ste
m
u
si
ng
m
achine learn
i
ng
a
ppr
oach as w
el
l as fo
c
us
in
g
on
data pri
vac
y
as seen
in
t
he work
of
Zh
ong et
al.
[
19]
and
Chen
et al
.
[
20
]
.
In
c
orp
or
at
io
n
of
t
he
sm
art
intel
le
ct
ual
syst
e
m
us
in
g
big
data
ap
proac
h
was
ca
rr
ie
d
out
by
Z
ha
ng
et
al
.
[2
1]
f
or
a
ssist
ing
ef
fecti
ve
dia
gnos
is.
S
tud
y
to
wards
E
HR
data
was
a
lso
carrie
d
out
by
Wu
et
al
.
[22]
as
well
as
Viceco
nti
et
al
.
[2
3]
wh
e
re
bi
g
data
appro
ac
h
has
been
us
e
d
for
m
on
it
or
in
g
the
healt
h
-
f
act
or
t
ow
a
r
ds
le
ading
t
o
an
eff
ect
ive
diag
nosis
syst
e
m
.
Stud
y
towa
r
ds
existi
ng
a
ppr
oa
ches
of
bi
g
da
ta
analy
ti
cs
t
ow
a
r
ds
i
m
pr
ovin
g
the
healt
hcar
e
sy
stem
is
carrie
d
out
by
Sh
a
f
qu
at
et
al
.
[24]
towa
rd
s
e
volvin
g
ou
t
wit
h
cost
eff
ect
ive
sc
he
m
es
of
existi
ng
tim
es.
A
la
ye
r
-
ba
sed
a
ppr
oach
was
car
r
ie
d
out
by
C
he
n
et
al
.
[25]
wh
e
re
a
sp
eci
fic
cas
e
stud
y
of
di
abetes
has
be
en
co
ns
ide
re
d
with
an
inte
ntion
for
ev
ol
ving
up
with
highly
custom
iz
ed
treatm
ent.
The
w
ork
of
Gar
at
ti
ni
et
al
.
[26]
ha
s
pr
e
sente
d
dis
c
us
sio
n
of
sig
nificat
or
at
trib
ute
of
us
in
g
bi
g
data
appro
ac
h
for
m
anag
in
g
dia
gnos
ti
c
analy
s
is
ov
e
r
le
thal
disease.
Util
izati
on
of
H
bas
e
for
m
edical
databa
se
m
anag
e
m
ent
is
carried
out
by
Ch
rim
es
a
nd
Zam
ani
[2
7]
wh
ere
bi
g
da
ta
analy
ti
cs
has
be
e
n
form
ulate
d
to
wards
c
ontrib
ut
ion
on
va
ried
cl
inica
l
serv
ic
e
s.
Ne
arly
,
the
si
m
il
ar
ty
pes
of
the
ap
proac
h
we
re
al
so
discuss
e
d by the
Tawal
be
h
et
al.
[
28
]
.
The
w
ork
of
S
ark
a
r
et
al
.
[29
]
has
pr
ese
nted
a
discuss
io
n
of
a
novel
sec
uri
ty
m
od
el
towar
ds
m
edical
dataset
.
The
w
or
k
of
Zha
ng
et
al
.
[3
0]
have
discuss
e
d
ab
ou
t
the
sig
nifi
cance
of
m
achine
le
arn
in
g
to
wards
sm
art
clinical
serv
ic
es
.
T
he
work
of
S
riniva
san
a
nd
A
run
asal
a
m
[3
1]
ha
s
discu
ssed
ab
ou
t
a
uniq
ue
f
or
m
of
data
an
al
yt
ic
al
syst
e
m
toward
s
em
ph
asi
zi
ng o
ve
r
t
he
offer
i
ng
s
ec
ur
in
g
bi
g
data
in
m
edi
cal
data. A
cl
ose
r
lo
ok
into
the
e
xisti
ng
syst
e
m
towards
tra
nsfo
r
m
at
ion
an
d
an
al
yt
ic
al
based
appr
oach
c
ons
iderin
g
m
edical
data
sh
ows that it
has adopti
on of
m
achine learn
i
ng
or
big
data appr
oach
wh
e
r
e the idea is to f
in
d
an
el
li
te
s
olu
ti
on
as
a
par
t
of
t
he
data
tran
sf
orm
at
ion
syst
e
m
.
The
a
ppr
oac
hes
are
cl
aim
e
d
to
offer
bette
r
outc
om
e
toward
s
addressi
ng
the
prob
le
m
s
sta
t
ed
in
their
re
s
pecti
ve
li
te
ratu
re.
Th
ey
existi
ng
wor
ks
are
associat
ed
with
both
adv
a
ntage
ous
f
act
s as w
el
l as
lim
it
i
ng
f
act
ors
too.
The
w
ork
of
Bhall
a
and
Ba
gg
a
[
32]
introdu
ce
s
a
m
od
el
known
as
op
inion
m
ining
f
or
the
te
xt
cat
egorizat
ion
.
This
m
od
el
ha
s
bee
n
dev
el
op
e
d
by
us
in
g
RB
-
bayes
m
od
el
.
T
he
RB
-
Ba
ye
s
com
pat
at
ion
i
s
hav
i
ng
acc
urac
y
83
.
34.
T
he
work
of
Aic
h
et
al
.
[33
]
introdu
ce
s
a
m
od
el
cal
le
d
Neural
-
netw
ork
to
cat
egorize
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Framew
or
k
for
eff
ic
ie
nt tran
sf
or
m
atio
n
f
or
c
omplex
med
ic
al
d
at
a
fo
r
im
provi
ng
analyti
cal
...
(
Ch
andru
AS
)
4855
the
we
b
-
base
d
te
xt.
This
m
odel
is
us
ef
ul
f
or r
esol
ving
the re
al
-
tim
e
pr
oble
m
fo
r
dif
fer
e
nt
ty
pes
of
te
xt m
ining
appr
oach
es
.
T
he
ne
xt
sect
io
n
outl
ines
the
pro
blem
that
a
re
extracte
d
af
te
r
rev
ie
wing
an
d
is
co
ns
i
de
red
as
the esse
ntial
r
e
seach
pro
blem
s to be s
olv
e
da
nd is a
ddresse
d i
n pro
po
se
d st
ud
y.
The pr
oble
m
s t
hat are
yet
uns
olv
e
d
in
leve
ra
ing
t
he
a
naly
ti
cal
p
r
ocessi
ng
of m
edical
d
at
a are:
a.
Existi
ng
stu
dies
are
s
pecific
to
disease
-
ba
s
ed
a
ppro
ac
h
w
hich
narr
ow
s
dow
n
the
s
co
pe
of
t
he
e
xisti
ng
appr
oach
es
to
wards
ge
ner
al
i
zed
diag
nosis
ov
e
r
m
edical
dat
a.
b.
Ther
e
is
no
wi
de
range
of
co
ns
ide
rati
on
to
t
he
fact
that
E
HR
data
are
now
ge
ne
rated
i
n
the
f
or
m
of
data
stream
w
her
e i
t i
s quit
e chall
eng
i
ng to
a
pply
o
nline
analy
ti
cs on t
he
to
p of
it
.
c.
The
gran
uliat
y
of
the
data
and
it
s
internal
structur
e
is
s
om
et
hin
g
wh
ic
h
is
sign
ific
an
tl
y
m
issi
ng
from
the ex
ist
in
g
sy
stem
that cause
s d
e
gr
a
datio
n
i
n
the
accu
racy
in the a
naly
ti
cal
p
r
ocess
.
d.
Ap
a
rt
f
r
om
fr
eq
ue
ntly
us
e
d
ET
L
sc
he
m
e,
there
is
no
m
uch
novelty
to
ward
s
le
ver
a
ging
data
trans
form
ation
schem
e tow
ar
ds faci
li
ta
ti
ng
be
tt
er knowled
ge
ex
tract
io
n p
r
ocess.
Ba
sed
on
the
above
uns
olv
e
d
pr
oble
m
fr
om
existi
ng
syst
em
,
the
fr
am
ing
of
the
pro
ble
m
sta
teme
nt
can
be
as
fo
ll
ows
“
Devel
opin
g
a
c
omp
uta
ti
onally
c
os
t
ef
fe
ct
iv
e
da
ta
tra
nsf
or
m
atio
ns
sc
he
me
for
facil
it
ating
pr
oc
essin
g
t
he c
omplex
med
ic
al data
for
b
et
t
er an
alyti
cal oper
ation.
”
The
Pro
po
se
d
So
luti
on
.
T
he
pr
im
e
aim
of
t
he
pro
posed
s
yst
e
m
is
to
evo
lve
up
an
ef
f
ect
ive
data
trans
for
m
ation
schem
e
that
i
s
capa
ble
of
r
end
e
rin
g
the
s
uitabil
it
y
of
th
e
com
plex
m
edical
data
in
orde
r
to
op
ti
m
iz
e
the
st
or
a
ge
as
well
as
le
ver
age
the
analy
ti
cal
pr
oc
essing
of
the
m
edical
data.
The
pro
posed
syst
e
m
adopts
an
ana
ly
ti
cal
research
m
e
tho
dolo
gy
in
or
de
r
to
achieve
it
s
a
i
m
.
The
pictorial
rep
rese
nta
ti
on
of
the m
et
ho
dolo
gy ado
pted
i
n pr
opos
e
d
s
ol
ution i
n sh
own
in
Fig
ur
e
1
.
The
pro
pose
d
syst
e
m
ta
kes
the
input
of
the
un
str
uctu
re
d
m
edical
data
wh
ic
h
le
ads
to
gen
e
rati
on
of
su
it
ably
struct
ur
e
d
data
in
a
non
-
co
nventio
nal
m
ann
er
in
order
t
o
m
ai
nt
ai
n
cost
ef
fecti
ven
e
ss
of
m
edical
data
analy
sis. Th
e
pro
posed
syst
e
m
co
ns
ide
rs
El
ect
ronic Healt
h R
ecord (
E
HR)
d
at
a whic
h
c
onsist
s of both
c
li
nical
as
well
as
non
-
cl
inica
l
infor
m
at
ion
about
the
patie
nt.
As
su
m
ing
that
a
hea
lt
hcar
e
syst
e
m
ho
sts
it
s
stora
ge
serv
ic
es
ove
r
cl
oud
as
well
as
ass
um
ing
that
healt
hcar
e
syst
em
ado
pt
s
va
rio
us
ad
va
nced
te
ch
nolo
gies
t
o
captu
red
aut
onom
ou
s
esse
ntial
EHR
i
nfor
m
at
ion
of
the
pa
ti
ent;
it
can
be
co
ns
ide
re
d
th
at
al
l
these
E
H
R
data
are
forw
a
r
ded
to
the
datace
nt
er
in
the
f
or
m
of
data
stream
wh
ic
h
is
f
ur
th
e
r
s
ubj
ect
e
d
to
stora
ge
fo
ll
ow
ed
by
analy
sis.
H
ow
ever,
diff
e
ren
t
from
any
existi
ng
syst
em
,
the
pro
posed
syst
e
m
con
si
der
s
that
ra
w
m
edical
dat
a
sh
oul
d
not
be
directl
y
store
d
in
cl
oud
st
or
a
ge
un
it
a
nd
in
ste
ad
a
pr
ocessi
ng
is
a
ppli
ed
on
the
to
p
of
it
to
m
ake
the
data
m
or
e
su
it
able
f
or
st
orage
a
nd
analy
sis.
In
t
his
re
ga
rd
s
,
the
pro
posed
syst
em
con
str
ucts
an
a
na
ly
ti
ca
l
env
i
ronm
ent
that
aggre
gates
a
ll
the
unstr
uctu
red
data
i
n
a
ve
ry
uniq
ue
for
m
wh
ic
h
is
f
urt
her
sub
j
ect
ed
t
o
next
par
t
of pr
ocess
ing
.
The
pr
opos
e
d
s
stem
ex
tract
s d
at
a chunk
s
(
or
ind
ivi
du
al
patie
nt d
at
a as a ce
ll
an
d
in
dex
it
s
v
al
ues
a
nd
al
l
the
cor
e
fiel
ds
(
al
so
know
n
as
hea
ders)
fol
lowed
by
a
un
iqu
e
i
nd
e
xing
op
e
rati
on.
T
his
ind
e
xing
oper
at
i
on
assist
s
in
c
onne
ct
ing
al
l
the
r
especti
ve
valu
es
of
eac
h
c
ore
fiel
ds
in
a
te
m
po
rar
y
bu
ff
e
r
ov
e
r
cl
oud
w
hile
al
l
the
sta
ti
c
header
inf
or
m
at
ion
are
store
d
pe
r
m
anen
tl
y
in
the
cl
ou
d
st
or
a
ge
un
it
s.
T
he
be
nf
ic
ia
l
facto
r
of
this
process
is
to
r
esi
st
the
it
erati
ve
ide
ntific
at
io
n
of
c
or
e
fiel
ds
an
d
sto
rin
g
the
sam
e
wh
ic
h
consi
der
a
bly
saves
the
processi
ng
effor
t
an
d
ti
m
e
.
The
pro
po
se
d
syst
e
m
al
so
us
es
a
virtu
al
te
m
pla
te
based
a
ppr
oach
i
n
ord
er
to
facil
it
at
e the p
r
ocess o
f
in
de
xin
g an
d data st
r
uctu
rizat
ion
usi
ng
a
m
at
rix
-
ba
sed o
per
at
io
n.
The
pro
posed
syst
e
m
us
es
a
synta
ct
ic
al
pa
rsing
wh
e
re
s
e
m
antic
-
base
d
ap
proac
h
is
a
pp
li
ed
with
higher
de
gree
of
custom
iz
at
ion
over
the
values
of
th
e
cor
e
fiel
ds
.
The
un
st
ru
ct
ur
e
d
data
is
t
her
e
by
trans
form
ed
to
sem
i
-
structur
e
d
a
nd
fi
n
al
ly
to
str
uctu
re
d
m
edical
data
where
fi
nally
sem
antic
s
are
a
ppli
ed
t
o
extract
t
he
c
orrect
in
fer
e
nce
of
the
m
edical
data
in
the
f
orm
of
kn
ow
le
dge.
The
nex
t
s
ect
ion
discu
sse
s
about
the im
ple
m
ent
at
ion
of prop
ose
d
syst
em
w
it
h
res
pect to al
go
rithm
an
d
it
s r
e
sp
ect
ive e
xec
ution
.
U
n
s
t
r
u
c
t
u
r
e
d
D
a
t
a
V
i
r
t
u
a
l
T
e
m
p
l
a
t
e
S
t
r
e
a
m
D
a
t
a
C
h
u
n
k
C
o
r
e
F
i
e
l
d
s
v
a
l
u
e
s
D
a
t
a
S
t
r
u
c
t
u
r
i
z
a
t
i
o
n
S
y
n
t
a
c
t
i
c
a
l
P
a
r
s
i
n
g
K
n
o
w
l
e
d
g
e
E
x
t
r
a
c
t
i
o
n
I
n
d
e
x
i
n
g
M
a
t
r
i
x
-
B
a
s
e
d
Figure
1
.
A
dopt
ed
m
et
ho
dolo
gy for
p
r
opos
e
d
tra
nsfo
rm
at
io
n
a
nd
a
naly
ti
cal
s
yst
e
m
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
53
-
48
62
4856
2.
SY
STE
M
I
M
PLE
MENT
A
TION
The
pro
po
se
d
syst
e
m
intro
du
ces
a
uniq
ue
f
or
m
of
sm
art
EHR
syst
em
t
hat
is
capa
ble
of
sto
rin
g
the
m
ined
in
form
ation
witi
n
t
he
cl
oud
distri
bu
te
d
st
or
a
ge
s
yst
e
m
insp
it
ed
of
st
or
in
g
the
raw
m
edical
data
in
it
.
Hen
ce
the
ta
rg
et
is
to
ob
ta
in
highe
r
de
gr
ee
of
st
or
a
ge
optim
iz
at
ion
an
d
assist
in
g
a
sm
oo
th
a
na
ly
ti
ca
l
op
e
rati
on
f
or
the
EHR
da
ta
.
This
sect
ion
discuss
es
a
bout
the
assum
ption
and
de
pe
nden
ci
es
associat
ed
wit
h
the d
e
sig
n
c
onstructio
n fo
ll
owed
b
y i
m
plem
entat
ion
strat
egy an
d
e
xec
ut
ion
flo
w.
2.1
.
Assu
m
pt
i
on
and
de
pende
nc
ie
s
The
pr
i
ma
ry
assu
mp
ti
on
of
the
pro
posed
sy
stem
is
that
there
are
var
i
ous
nu
m
ber
s
of
he
at
hcar
e
unit
s
wh
ic
h
has
a
n
auto
no
m
ou
s
m
echan
ism
to
up
loa
d
the
EH
R
data
wh
ic
h
are
fr
ee
from
any
error
s
or
a
rtifact
s
.
It
will
m
ean
that
there
are
no
pro
blem
s
within
the
EHR
data
from
the
local
so
ur
ce
of
ori
gina
ti
on
.
The
sec
ondar
y
as
s
umptio
n
of
t
he
pr
opose
d
s
yst
em
is
t
hat
al
l
these
healt
hcar
e
c
ha
ins
are
c
onne
ct
ed
to
cl
oud
-
base
d
a
serv
ic
e
w
hic
h
extracts
the
div
ersi
fied
E
HR
data
from
m
ulti
ple
so
urces
and
respo
sit
it
in
distrib
uted
ord
er
in
the
cl
ou
d
cl
us
te
rs
.
Th
e
te
rti
ar
y
as
sumptio
n
of
the
propose
d
syst
e
m
is
th
at
there
are
div
e
rse
f
or
m
s
of
E
HR d
at
a
be
ing
c
ollec
te
d
and
when
t
he
pro
blem
s
agg
re
gated
or
a
rtifac
ts
are
intr
oduc
ed
ove
r
a
cl
ou
d
respos
it
ion
process
.
Ther
e
f
or
e,
th
e
go
al
of
the
pr
opos
e
d
syst
em
will
be
to
address
this
pro
bl
e
m
by
introd
ucin
g
a
te
chn
i
qu
e
t
hat
can
recti
fy
this
data
integrati
on
pro
blem
in
c
loud
in
ord
er
to
re
du
ce
t
he
adv
e
rs
e
eff
ect
of unstr
uc
ture
d
data. T
he
core d
e
pende
ncies
of
t
he
pr
opos
e
d
syst
em
is al
l
the EH
R data i
s in
the form
at
of
te
xt
a
nd
co
uld
be
res
posi
te
d
in
any
fo
r
m
at
;
ho
we
ve
r,
fo
r
ea
sines
s
in
com
pu
ti
ng,
the
pro
posed
s
yst
e
m
consi
ders the
text is i
n plai
n t
ext form
at
.
2.2
.
Implem
ent
ati
on
s
tr
at
e
gy
It
shou
l
d
be
no
te
d
that
pr
opos
e
d
syst
em
is
a
co
m
pu
ta
ti
on
al
m
odel
that
ta
rg
et
s
to
optim
iz
e
the
distrib
uted
stora
ge
syst
e
m
fo
r
healt
hcar
e
un
it
in
order
t
o
facil
it
at
e
s
mart
analy
ti
cal
o
per
at
io
ns
.
T
he
stud
y
of
t
he
pro
pose
d
syst
em
de
m
ands
a
stream
of
i
nco
m
ing
data
from
the
so
urce
of
t
he
EHR
data;
howev
e
r,
su
c
h
m
echan
ism
is
qu
it
e
a
com
plex
Ther
ef
ore
,
the
pro
po
se
d
syst
em
con
structs
a
n
analy
ti
cal
data
aggre
gatio
n
m
od
el
as
the
pr
im
ary
i
m
p
leme
ntati
on
strat
egy
so
that
ne
ar
-
ti
m
e
traff
ic
flo
w
of
te
xt
ual
EHR
data
cou
ld
be
m
app
ed
i
n
pro
po
s
ed
syst
em
.
Af
te
r
t
he
proce
ss
of
data
ag
gregati
on
is
a
naly
ti
cal
ly
design,
the
ne
xt
proces
s
wil
l
be
to
pe
rfo
rm
transfo
rm
ation
operati
on
f
r
om
un
struct
ured
data
to
se
m
i
-
structur
e
d
data
so
that
a
naly
ti
cal
op
e
rati
on
ca
n
be
act
ually
car
ried
out
on
the
to
p
of
it
.
T
he
nex
t
pa
rt
of
th
e
stu
dy
im
ple
m
ents
a
c
on
te
xt
-
base
d
m
echan
ism
whic
h
ex
t
racts
the
m
ined
in
form
at
ion
on
the
ba
sis
of
the
c
orre
la
te
d
data
ob
ta
ined
from
se
m
antic
s
of
t
he
te
xt
extr
act
ed.
F
or
this
pur
po
se
,
the
pr
opos
e
d
syst
em
perf
or
m
s
init
i
al
cl
assifi
cat
ion
of
t
he
te
xt
da
ta
int
o
two
ty
pes
viz.
i)
sta
ti
c
data
and
ii
)
dy
nam
i
c
us
e
r
d
at
a
.
F
or
faster
proce
ssing,
the
pro
pose
d
syst
em
e
xtracts
the
sta
ti
c
data
and
re
posit
over
the
cl
ou
d
stora
ge
unit
s
wh
il
e
f
or
m
ain
intai
ning
non
-
re
dundant
dat
a
ov
e
r
the
distri
bute
d
cl
oud
st
or
a
ge
s
yst
e
m
,
the
pro
po
s
ed
syst
em
e
xtracts
t
he
dynam
ic
data,
in
de
x
i
t
a
nd
the
n
re
po
sit
ov
e
r
t
he
te
m
po
ra
ry
bu
ff
e
r.
The
pro
posed
al
gorithm
is
t
hen
ap
plied
to
this
te
m
po
rary
buff
e
r
i
n
order
t
o
facil
it
at
e
data
trans
form
ation
op
e
rati
on
fo
ll
ow
e
d
by
e
xtra
ct
ion
of
m
ining
ap
proac
h.
I
n
the
entire
pro
cess,
the
pro
posed
s
yst
e
m
of
fer
s
si
gn
i
ficant
am
ou
nt
of
cost
e
ff
e
ct
ive
analy
ti
cal
op
e
rati
on
ov
er
the
m
edical
data.
The
e
xec
ution
process
is as s
how
n
i
n nex
t
sub
-
sect
io
n.
2.3
.
Executi
on
flo
w
The
fi
rst
ste
p
of
the
e
xecu
ti
on
flo
w
is
to
pro
cess
the
in
put
stream
of
te
xt
data.
H
owev
er,
it
is
pr
io
rly
discusse
d
t
hat
extracti
on
of
s
tream
s
of
te
xt
data
is
chall
en
ging
a
nd
nee
ds
an
acce
s
s
to
the
bu
ff
e
r
of
cl
oud
stora
ge
points
wh
ic
h
m
ai
ntain
s
an
a
da
ptive
qu
e
ue
syst
em
.
Howe
ver,
the
pro
po
se
d
syst
e
m
at
tem
pts
to
reali
ze
this p
roblem
b
y const
ru
ct
in
g i
ts o
w
n b
uffer
of stream
d
at
a d
s
ov
e
r
a
sam
pl
ed peri
od of ti
m
e
t
s,
see
Fig
ur
e
2
.
-
-
a
1
a
2
-
-
a
1
a
2
-
-
a
1
a
2
d
i
1
(
d
s
1
)
d
i
2
(
d
s
2
)
d
i
n
(
d
s
n
)
t
s
d
a
d
m
e
m
Fi
gure
2
.
Str
uc
ture of
inc
om
i
ng d
at
a
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Framew
or
k
for
eff
ic
ie
nt tran
sf
or
m
atio
n
f
or
c
omplex
med
ic
al
d
at
a
fo
r
im
provi
ng
analyti
cal
...
(
Ch
andru
AS
)
4857
Ther
e
f
or
e,
the
first
ste
p
of
th
e
pro
pose
d
e
xe
cution
is
t
o
c
onstr
uct
a
synth
et
ic
un
st
ru
ct
ured
data
ov
e
r
the
pro
posed
buf
fer
syst
em
.
The
al
gorithm
tak
es
the
in
put
of
data
stream
ds
ov
e
r
a
def
init
e
sam
pled
per
iod
of
tim
e ts wh
ic
h
a
fter
processi
ng yi
el
ds
tem
_
m
a
trix (
te
m
po
ra
ry
m
a
tris of u
ns
tr
uctu
red data
).
Algorithm for Synthetic Unstructured Data Construction
Inpu
t
: d
s
(data stream)
Output
: tem_matrix (temporary matris of unstructured data)
Start
1.
For
i=1:t
s
2. d
i
ex
tr
ac
t(
d
s
)
3. d
a
ag
g(
d
i
)
4. d
mem
r
ep
os
it
(d
a
)
5.
For
j=1:
a
6. extract d
stat
(
h
, p
o
)
7. reposit d
stat
cloud_templ
ate
8. extract d
dyn
9. reposit d
dyn
tem_matrix
10.
End
11.
End
End
The
discuss
i
on
s
of
ste
ps
of
t
he
al
gorithm
are
as
f
ollo
ws:
The
al
go
rithm
consi
ders
that
there
is
a
n
ind
ivi
du
al
data
d
i
f
or
eac
h
res
pecti
ve
stream
of
data
d
s
(
Line
-
1).
T
he
syst
e
m
than
ag
gr
e
gates
al
l
the
sam
pled
incom
ing
strea
m
d
s
and
sto
r
e
it
in
an
aggreg
at
e
d
m
at
rix
d
a
(Line
-
2)
.
All
these
data
are
the
cum
ulati
vely
aggre
gated
fo
ll
ow
e
d
by
re
po
s
it
ing
da
i
n
ord
er
to
f
orm
a
sa
m
pled
bu
f
fer
d
m
e
m
(Line
-
4).
T
her
e
fore,
t
he
buf
fe
r
dm
e
m
can
be
con
si
der
e
d
to posses
al
l t
he
agg
re
gated
strea
m
s
o
f
sa
m
pled
data d
a
, which c
an
be
no
w
subj
ect
e
d
to
further
pro
cessi
ng.
The
pro
po
se
d
syst
em
con
siders
t
hat
each
ind
i
vi
du
al
data
di
consi
sts
of
in
div
id
ual
patie
nt
E
HR
r
ecords
a
,
w
he
re
the
va
riabl
e
a
will
nu
m
ber
of
uni
qu
e
EHR
recor
d
of
i
nd
i
vidual
patie
nt.
Ther
e
a
re
good
po
s
sibil
it
ie
s
that
nu
m
ber
of
a
in
one
in
div
id
ual
data
d
1
c
ou
ld
dif
fer
i
n
an
ot
her
in
div
i
du
al
d
2
as
well
as they c
ould als
o be sa
m
e, it al
l dep
e
nd
s
up
on the
densit
y of t
he
te
x
t data i
n
t
he
tr
aff
ic
.
Ther
e
f
or
e,
co
nsi
der
in
g
al
l
t
he
value
s o
f
a
(Li
ne
-
5),
the
al
go
rithm
extracts
sta
ti
c
data
d
stat
with r
espect
to
the
hea
de
rs
he
ad
a
nd
poi
nters
p
o
(Line
-
6).
Ba
sic
al
ly
,
hea
ders
a
re
t
he
pr
im
e
at
tribu
te
of
t
he
fiel
d
that
represe
nts
the
com
plete
colum
n
of
te
xt
data
with
resp
ect
t
o
it
s
ty
pe
w
hil
e
pointer
c
onne
ct
s
al
l
the
hea
der
wit
h
it
s
resp
ect
ive
value
d
d
y
n
.
It
will
m
ean
that
pro
po
se
d
syst
e
m
m
ake
us
e
of
a
te
m
plate
base
d
data
reposit
io
n
te
chn
iq
ue
w
he
re
ds
ta
t
repres
ents
sta
ti
c
data
ov
er
the
te
m
plate
wh
il
e
d
d
y
n
represe
nts
patie
nt
EHR
data
fed
by
the
per
s
onel
of
the
healt
hcare
sect
or
.
It
interp
rets
that
d
d
y
n
is
a
direct
value
to
represen
t
ind
ivid
ual
he
aders
(h
1
,
h
2
,
…
)
w
her
e
pointers
s
egr
e
gates
header
with
i
nd
i
vid
ual
value
of
EHR
(Li
ne
-
7
and
Lin
e
-
8).
F
inall
y,
al
l
the
ds
ta
t
data
are
respos
it
ed
ov
e
r
cl
oud
-
te
m
plate
storag
e
wh
il
e
use
r
EHR
data
i.e.
d
d
y
n
is
s
tored
in
a
tem
po
ra
ry
m
at
rix
te
m_m
atrix
(Li
ne
-
9)
w
hich
has
al
l
the
unstr
uct
ur
e
d
data.
T
he
co
ns
trib
utio
n
of
t
his
al
gorithm
is t
hat it
of
fe
rs
a sa
ndboxi
ng
m
ec
han
ism
to
con
s
truct a
m
e
m
or
y sy
stem wh
ere
the u
nst
ru
ct
ur
e
d
data
can
be
sto
re
d
in
order
t
o
facil
it
at
e
u
pco
m
ing
processi
ng
of
data.
Fi
g
ure
3
pictorial
ly
repr
esents
the
proc
essin
g
of the
data stre
a
m
an
d f
ollowe
d by re
posit
ing
it
.
-
-
a
1
a
2
-
-
a
1
a
2
-
-
a
1
a
2
d
i
1
(
d
s
1
)
d
i
2
(
d
s
2
)
d
i
n
(
d
s
n
)
t
s
d
a
d
m
e
m
d
s
t
a
t
d
d
y
n
S
t
o
r
e
d
i
n
c
l
o
u
d
S
t
o
r
e
d
i
n
t
e
m
_
m
a
t
r
i
x
i
n
d
e
x
i
n
g
Figure
3
.
Me
ch
anism
o
f
r
e
pos
it
ing
d
at
a
s
trea
m
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
53
-
48
62
4858
The
ne
xt p
a
rt of the
im
ple
m
entat
ion
is asso
ci
at
ed
with the e
xtracti
ng know
le
dg
e f
r
om
the d
at
a stor
e
d
in
tem
po
ra
ry
m
at
rix.
The
al
gorithm
ta
kes
the
input
of
h
(
hea
der
s
)
an
d
d
d
y
n
(d
yna
m
ic
data)
w
hich
after
processi
ng yi
el
ds
a
nd
ou
tc
ome
of
d
sem
_str
uc
(se
m
i
-
structured
data) a
nd term
k
now
(
m
ined
data
).
Algorithm for Mined Data Extraction using Semantic
Input
:
h
(headers), d
dyn
(dynamic data)
Output
: d
sem_struc
(semi
-
structured data), term
know
(mined data)
1.
For
h=1:max
2. t1
ad
d
<
a
nd
>
t
ag
s
to
h
and d
dyn
3. reposit in d
sem_struc
4. [term loc]
se
ma
nt
i
c(
t1
)
5. extract term
know
te
rm
6.
End
The
disc
us
sio
ns
of
ste
ps
of
t
he
al
gorithm
a
re
as
fo
ll
ows:
In
order
to
perfo
m
ing
m
ining
op
e
rati
on,
it
is
necessary
for
the
al
gorithm
fo
r
e
ns
ure
pro
per
c
onfirm
at
ion
of
the
use
r
EHR
rec
ord
s.
As
t
her
e
a
re
al
so
po
s
sibil
it
ie
s
of
re
dundant
dat
a
(
wh
ic
h
m
ay
aff
ect
t
he
fu
t
ure
pa
rt
of
a
naly
ti
cal
op
erati
on),
t
her
e
is
a
ne
ed
t
o
perform
a
pro
per
in
dex
i
ng
op
erati
on.
F
or
this
purpose
,
t
he
al
gorithm
first
acce
sses
the
te
m
po
ra
ry
m
at
rix
wh
ic
h
has
al
l
the
us
e
r
-
fed
E
HR
recor
ds
of
the
data;
howe
ver,
there
is
no
sta
ti
c
data
in
the
tem
po
ra
ry
m
at
rix.
Ther
e
f
or
e,
t
he
al
go
rithm
will
need
to
c
hec
k
the
in
de
xes
of
the
eac
h
h
e
ader
s
pr
e
sent
i
n
the
va
riable
a
an
d
const
ru
ct
a
nd
ind
e
x.
T
he
const
ru
ct
io
n
of
the
in
de
x
is
carrie
d
out
on
t
he
basis
of
t
he
te
rm
us
ed
in
the in
div
id
ual
head
e
rs wit
h re
sp
ect
to
it
s in
div
id
ual locati
on
.
The
pr
opos
e
d
syst
e
m
cal
cula
te
s
the
nu
m
ber
of
e
ntries
f
or
each
in
div
id
ua
l
sa
m
pled
data
stream
s
in
order
to
init
ia
ll
y
con
fi
rm
the
owne
rsh
i
p
of
respec
t
ddyn
for
the
res
pe
ct
ive
ind
i
vidu
al
head
e
rs
h
.
Fo
r
a
n
exam
ple,
consi
der
t
hat
the
re
are
th
ree
headers,
i
n
s
uc
h
c
a
se,
h
1(loc1)
d
d
y
n1
,
h
2(loc2)
d
d
y
n
2
,
an
d
h
3(loc3)
d
d
y
n
3
with
resp
ect
t
o
three
dif
fer
e
nt
locat
ions
loc
1
,
loc
2
,
a
nd
loc
3
resp
ect
ively
.
T
his
in
de
x
of
t
he
res
pecti
ve
he
ader
s
and
l
ocati
on
s
are
store
d
in
c
loud
unit
and
is
resp
ect
ively
ind
e
xed
with
the
res
pect
ive
us
er
fed
E
HR
data.
The
al
gorithm
init
ia
ll
y
con
sid
ers
al
l
the
m
axi
m
u
m
max
hea
der
s
pr
ese
nt
in
data
stream
s,
fo
ll
owe
d
by
ad
ding
sta
rt
ta
g
<
an
d
end
ta
g
/>
(Li
ne
-
2).
T
his
operati
on
is
car
ried
ou
t
for
he
ader
h
a
nd
us
e
r
fe
d
E
HR
dat
a
d
d
y
n
,
wh
ic
h
res
ults
in
sem
i
-
structure
data
stored
i
n
m
a
trix
d
sem
_st
ruc
fr
om
un
struc
ture
d
data
(Li
ne
-
3).
It
shou
l
d
be
no
te
d
that
t
his
op
e
rati
on
is
ca
rr
ie
d
out
wtho
ut
us
i
ng
a
ny
e
xisti
ng
to
ols
of
any
distrib
uted
f
ram
ewo
r
k
a
nd
ye
t
they
are q
uite f
ast
er and
e
ff
ic
i
ent in t
heir usa
ge.
The
nex
t
par
t
of
t
he
pr
opose
d
syst
em
is
about
usi
ng
the
sem
antic
op
erati
on
in
or
der
t
o
obta
in
the
co
rr
ect
e
d
m
eaning
a
nd
c
on
te
xt
of
the
m
edical
EHR
data
(Li
ne
-
4).
This
op
e
rati
on
al
lows
t
he
propose
d
syst
e
m
to
cons
truct
c
us
tom
ized
nu
m
ber
of
s
e
m
antic
s
on
th
e
basis
of
the
pro
blem
s
in
diagnosis
a
nd
hence
this
op
e
rati
on
with
m
ino
r
finet
un
i
ng
in
t
he
sem
a
ntics
can
res
ult
in
highly
i
m
p
rove
sco
pe
of
autom
at
ed
diagnosis
and
kn
ow
le
dg
e
extracti
on
pro
cess
.
T
he
pro
pose
d
syst
em
ca
n
c
on
st
ru
ct
s
se
m
antic
s
on
t
he
basis
of
t
he
na
m
e
of
the
prom
inent
disease
or
t
he
hi
gh
ly
sig
nificant
re
vie
w
of
the
dis
ease
m
ade
by
certai
n
ph
y
sic
ia
n.
The
c
on
st
ru
ct
e
d
sem
antic
s
are
app
li
ed
over
the
te
rm
s
t
1
fr
om
Line
-
2
w
hich
c
onnects
bo
t
h
sem
i
-
structur
e
d
head
e
rs
a
nd
use
r
fe
d
val
ue
of
EHR r
ec
ord of i
nd
ivi
du
al
patie
nt w
it
h resp
ec
t t
o
it
s ter
m
s a
nd
l
ocati
on
(Li
ne
-
4)
.
Finall
y,
the
e
xr
act
e
d
te
rm
is
co
ns
ide
red
to
be
fi
nal
out
com
e
of
the
pro
posed
al
gori
thm
in
the
for
m
of
knowle
dge.
Th
e
F
ig
ur
e
4 intr
oduces
the c
om
ple
te
p
r
ocess
of
propose
d
sy
stem
.
1
1
1
2
3
4
5
6
7
s
1
s
2
s
3
s
n
c
1
c
2
c
i
c
j
c
k
c
z
.
.
.
(
i
+
j
+
k
+
…
.
+
m
)
=
n
U
n
s
t
r
u
c
t
_
d
a
t
a
U
n
s
t
r
u
c
t
_
d
a
t
a
U
n
s
t
r
u
c
t
_
d
a
t
a
E
H
R
_
S
e
r
v
1
E
H
R
_
S
e
r
v
2
E
H
R
_
S
e
r
v
3
E
H
R
_
c
l
o
u
d
A
l
g
o
r
i
t
h
m
E
H
R
_
c
l
o
u
d
_
c
l
u
s
t
e
r
<
S
t
r
u
c
t
_
d
a
t
a
>
a
n
a
l
y
s
t
Figure
4
.
Com
plete
p
ro
ces
s
of
p
r
opos
e
d
s
yst
e
m
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Framew
or
k
for
eff
ic
ie
nt tran
sf
or
m
atio
n
f
or
c
omplex
med
ic
al
d
at
a
fo
r
im
provi
ng
analyti
cal
...
(
Ch
andru
AS
)
4859
3.
RESU
LT
A
N
ALYSIS
In
orde
r
to
ca
r
ry
out
analy
sis
of
t
he
propose
d
syst
em
,
the
chall
eng
i
ng
ste
p
is
the
pri
m
ary
process
of
ta
kin
g
a
pr
eci
s
e
input
of
E
HR
data.
This
is
because
a
c
omplex
form
of
E
HR
data
is
re
quire
d
f
or
this
pur
pose
wh
e
re
the
dat
a
is
la
rg
er
i
n
dim
ension
as
well
as
there
are
va
rio
us
form
s
of
heter
ogene
uity
in
th
e
data.
Ther
e
f
or
e,
the
pro
posed
syst
e
m
rev
ie
we
d
so
m
e
of
the
publica
ll
y
av
ai
la
ble
dataset
[34
-
37]
in
orde
r
t
o
visu
a
li
ze
the
pa
tt
ern
s
of
poss
ible
m
edical
data
in
the
big
data.
He
nce,
the
first
pr
oce
ss
is
to
ens
ur
e
that
pro
po
se
d
syst
e
m
a
lso
co
ns
id
er
E
HR
da
ta
wh
ic
h
is
sim
ilar
to
the
patte
r
n
of
bi
g
da
ta
.
The
st
ud
y
c
on
sidere
d
the
EHR
data
from
the
existi
ng
sta
nda
rd
da
ta
set
of
[36
,
37]
wh
e
re
each
file
s
are
of
siz
e
of
m
egab
yt
es
with
diff
e
re
nt
acce
ssible
form
at
s
t
hat
are
wi
dely
su
pp
or
te
d
on
any
m
achine.
Howev
e
r,
s
uc
h
avail
able
da
ta
set
do
e
sn’t
ha
ve
consi
ste
ncies
and
he
nce,
t
he
re
is
a
nee
d
to
co
ns
tr
uct
an
arti
fici
al
E
HR
data
a
nd
hen
ce
,
the p
r
opos
e
d
s
yst
e
m
co
ns
truc
ts a syntheti
c E
HR d
at
a whic
h con
sist
s of m
ulti
ple h
eaders wit
h
al
l head
er
s w
it
h
diff
e
re
nt
val
ue
s
in
orde
r
to
a
ssess
the
ef
fec
ti
ven
ess
of
m
i
ning
a
ppr
oach.
The
pr
opos
e
d
dataset
c
onsist
s
of
discrete
inf
orm
at
ion
about
the
patie
nts
as
well
as
it
a
lso
has
va
rio
us
cl
inica
l
infer
e
nce
s
fr
om
the
ph
ysi
cal
of
the
res
pecti
ve
patie
nt.
T
he
pur
pose
of
the
a
naly
sis
will
be
to
obta
in
a
nal
yt
ic
al
info
rm
ation
a
bout
the
disease
crit
ic
al
ity of
t
he
r
es
pecti
ve pa
ti
ent.
3.1.
Assessme
nt
en
vironm
ent
The
sc
riptin
g
of
the
pro
po
sed
syst
em
has
bee
n
ca
rr
ie
d
ou
t
i
n
M
A
TLAB
c
onsid
erin
g
norm
al
windows
e
nv
ir
on
m
ent
with
cor
e
-
i3
m
achine.
Ado
ption
of
MATLAB
off
ers
va
rio
us
be
nef
it
s
of
carryi
ng
out
trans
form
ation
op
erati
on
usi
ng
the
m
at
rix
-
base
d
m
echan
ism
in
or
de
r
t
o
assess
the
pr
op
os
e
d
syst
e
m
.
The
c
om
plete
i
nput
data
is
spl
it
te
d
into
sm
aller
ve
rsions
to
gen
e
rate
in
div
i
du
al
data
d
i
w
her
e
t
he
siz
e
of
one
ind
ivi
du
al
d
at
a
m
ay
diff
er from
each
oth
er. The idea is also to
assess the o
ve
rall
d
at
a p
r
ocessin
g
tim
e t
o
see if
diff
e
re
nt
file
s
iz
e
has
a
ny
ef
fect
over
the
t
hro
ughput
of
a
naly
ti
cal
op
era
ti
on
.
The
anal
ysi
s
of
t
he
pro
po
s
ed
stud
y i
s ca
rr
ie
d o
ut consi
der
in
g
ti
m
e fo
r data
structu
rizat
ion
and m
e
m
or
y sat
ur
at
io
n
sta
te
.
3.2.
Results
obt
ain
ed
The
asse
ssm
ent
of
pro
po
s
ed
s
yst
e
m
is
carried
out
for
ove
rvi
ewing
the
in
di
vidual
outc
om
e
as
well
as
com
par
at
ive
outc
om
e.
The
com
pu
ta
ti
on
of
the
tim
e
fo
r
data
struct
ur
i
zat
ion
is
carri
ed
out
by
obt
ai
ning
the
di
ff
e
ren
ce
betwee
n
the
ti
m
es
for
pr
e
vious
recor
d
with
the
cu
rr
e
nt
dat
a
struct
ur
iz
at
io
n
record
.
T
he
a
naly
sis
is
carried
out
ov
e
r
10
e
xp
e
ri
m
ental
tria
ls,
wh
e
re
the
al
lo
cat
ion
of
in
put
datast
ream
is
increase
d
ra
nd
om
l
y
in
order t
o
m
ap
w
it
h
the
near
-
rea
l world
d
at
a st
r
ea
m
ing
pr
oces
s over t
he netw
ork.
The
outc
om
e
i
n
Fig
ur
e
5
sho
ws
that
data
str
uctu
rizat
ion
ti
m
e
increase
wi
th
increase
in
e
xp
e
rim
ental
tria
ls,
w
hich
is
ver
y
m
uch
natur
al
in
orde
r.
It
is
beca
us
e
e
ve
ry
ex
per
im
ental
s
tria
ls
hav
e
va
ryi
ng
al
lo
cat
ion
of
siz
e
of
data
strea
m
s
with
10
-
20%
def
i
niti
ve
increm
ent
an
d
rest
80
-
90
%
al
locat
ion
of
data
is
carried
ou
t
i
n
rand
om
fash
io
n.
A
n
inte
resti
ng
obse
rv
at
i
on
to
see
in
this
par
t
of
a
naly
sis
are
tw
o
f
olds
viz.
i)
t
he
pro
po
s
ed
syst
e
m
of
fer
s
a
gr
a
dual
incr
ease
in
tim
e,
wh
ic
h
is
quit
e
determ
inist
ic
in
orde
r
an
d
s
hows
that
it
offe
rs
a
sta
bili
zed
da
t
a
processin
g
c
apab
il
it
y,
an
d
ii
)
irres
pecti
ve
of
fluct
uating
siz
e
of
data
st
ream
s,
the
pro
po
s
ed
syst
e
m
of
fer
s
bette
r
tran
sf
orm
at
ion
perfor
m
ance
wh
e
re
abno
rm
al
traff
ic
increase
do
e
sn
’t
hav
e
a
si
gnific
an
t
eff
ect
to
wa
rd
s
the p
e
rfo
rm
ing
of
perform
ing
an
al
yt
ic
a
l op
er
at
ion
.
Figure
5
.
O
utc
om
e o
f
d
at
a
s
tr
uctu
rizat
ion
t
i
m
e
The
ne
xt
pa
rt
of
the
a
naly
sis
is
check
f
or
the
tim
e
con
sum
pt
ion
f
or
t
he
cor
e
a
naly
ti
cal
pr
oce
ss
as
well
as
data
s
tructu
rizat
ion
process
.
Ba
sci
cal
ly
,
an
eff
e
ct
ive
data
str
uc
turizat
ion
al
ways
le
ad
to
highe
r
accuracy
of
th
e
m
ined
data;
howe
ver,
high
er
value
o
f
the
know
le
dg
e
e
xt
racti
on
tim
e
c
an
be
ex
pecte
d.
It
is
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
53
-
48
62
4860
because
after
the
data
struct
uri
zat
ion
is
carr
ie
d
ou
t
tha
n
the
syst
e
m
will
be
al
ways
require
d
to
cro
ss
chec
k
with
the
ind
e
xes
from
the
cl
ou
d
res
our
ces
in
or
de
r
to
co
nf
irm
the
po
sit
ion
a
nd
te
r
m
ob
ta
ined
from
the
dynam
ic
data.
This
co
ul
d
possi
bly
increase
the
knowle
dge
extr
act
ion
ti
m
e
to
som
e
extent.
Fig
ure
6
highli
gh
ts
t
he
sam
e
fact
wh
ere
it
is
sh
ow
n
that
knowle
dge
extracti
on
tim
e
is
sli
gh
tl
y
m
or
e
t
han
t
he
ti
m
e
for
daa str
uctu
rizat
ion
.
Figure
6
.
An
al
ysi
s o
f
o
ver
al
l
c
om
pu
ta
ti
on
al
t
i
m
e
Ap
a
rt
f
ro
m
t
his,
the
pro
pose
d
syst
em
has
been
c
om
par
ed
with
the
existi
ng
ETL
sch
em
e
(Ex
t
ract/
Transf
or
m
/Load)
w
hi
ch
is
the
f
re
qu
e
ntly
us
e
d
data
trans
f
or
m
at
ion
sc
hem
e
in
existi
ng
syst
e
m
(e.
g.
[11
-
13]
).
The
com
parat
ive
analy
sis
is
car
ried
out
with
re
sp
e
ct
to
re
spo
n
s
e
tim
e
and
m
e
m
or
y
consum
ption
a
s sho
wn in
Fig
ur
e
7
.
(a)
(
b
)
Figure
7
.
Com
par
at
ive
a
naly
s
is
,
(a
) An
al
ysi
s
of
r
e
spo
ns
e
t
im
e
,
(b) Analy
s
is of
m
e
m
or
y
c
on
s
um
ption
A
cl
os
er
lo
ok
into
the
outc
o
m
e
of
the
com
par
at
ive
analy
sis
in
Fig
ure
7
s
hows
th
at
propo
sed
syst
em
offer
s
sig
nifica
ntly
b
et
te
r per
f
or
m
ance as c
om
par
ed
to t
he e
xisti
ng syst
em
in term
s o
f
r
esp
on
se
tim
e
as sho
wn
in
Fig
ur
e
7(a
)
and
m
e
m
or
y
consum
ption
t
i
m
e
as
sh
own
in
Fig
ur
e
7(b
).
T
he
pri
m
e
r
easo
n
be
hind
t
his
is
ETL
sc
hem
e
i
s
shro
uded
wit
h
a
pro
blem
of
pa
rsing
with
the
s
ource
da
ta
especial
ly
of
the
data
is
m
assive
and
heter
og
e
ne
ous
in
it
sel
f.
Howev
e
r,
propose
d
syst
em
of
fe
rs
a
ve
ry
li
gh
twei
ght
schem
e
witho
ut
any
dep
e
ndencies
of
t
hir
d
par
ty
app
li
cat
io
ns
or
plugin
s
as
w
el
l
as
us
es
a
s
i
m
plifie
d
sem
antic
s
f
or
extr
act
ing
the
knowle
dge
.
The
outc
om
e
of
m
e
m
or
y
con
su
m
ption
al
s
o
sho
ws
that
pro
ps
e
d
syst
e
m
occupies
a
ve
r
y
low
a
m
ou
nt
of
m
e
m
or
y.
The
c
ore
reas
on b
ehi
nd
this
is
that p
r
opos
e
d
syst
em
us
es
a
sa
ndboxi
ng
m
echan
ism
w
here
a
tem
po
rar
y
buff
e
r
is
co
ns
tr
uc
te
d
w
hich
ca
n
sign
ific
a
ntly
save
the
c
onsu
m
ption
of
the
m
e
m
or
y.
He
nce,
there
is
no
rec
ord
of
the
unde
rly
ing
process
w
hich
m
akes
the
analy
ti
cal
op
erati
on
faste
r
as
well
as
m
e
m
or
y
eff
ic
ie
nt.
Th
e
r
efore, t
he pr
op
os
e
d
syse
m
o
ffers a c
os
t e
ff
ec
ti
ve
data tra
nsf
or
m
at
ion
proc
e
ss for EHR
d
at
a.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Framew
or
k
for
eff
ic
ie
nt tran
sf
or
m
atio
n
f
or
c
omplex
med
ic
al
d
at
a
fo
r
im
provi
ng
analyti
cal
...
(
Ch
andru
AS
)
4861
4.
CONCL
US
I
O
N
T
he
pa
pe
r
p
r
e
s
e
nt
s
a
un
i
qu
e
a
pp
r
oa
c
h
o
f
t
r
a
ns
f
o
r
m
at
i
on
w
he
r
e
t
he
c
om
pl
e
x
un
s
t
r
uc
t
ur
e
d
s
t
r
e
am
of
m
e
di
c
al
da
t
e
i
s
c
on
ve
r
t
e
d
t
o
h
i
gh
l
y
s
t
r
uc
t
ur
e
d
da
t
a
i
n
di
f
f
e
r
e
nt
m
ec
ha
ni
sm
t
ha
t
i
s
f
r
e
e
f
r
om
a
ny
de
pe
nd
e
nc
i
e
s
.
T
he
c
o
nt
r
i
bu
t
i
on
of
t
he
pr
o
p
os
e
d
s
y
s
t
em
a
r
e
a
s
f
ol
l
o
w
:
i
)
t
he
p
r
o
po
s
e
d
s
y
s
t
em
r
e
du
c
e
d
t
he
s
t
or
a
ge
c
om
pl
e
xi
t
y
a
s
w
e
l
l
a
s
i
nc
r
e
a
s
e
s
f
a
s
t
e
r
pr
oc
e
s
s
i
ng
t
im
e
,
i
i
)
t
he
pr
op
os
e
d
s
y
s
t
em
i
nt
r
od
uc
e
s
a
no
ve
l
m
e
c
ha
ni
sm
t
o
i
de
nt
i
f
y
s
t
a
ti
c
a
nd
dy
na
m
i
c
da
t
a
w
he
r
e
t
he
s
t
a
t
i
c
da
t
a
i
s
s
t
or
e
d
i
n
c
l
ou
d
a
nd
dy
na
m
ic
da
t
a
i
s
s
t
or
e
d
i
n
t
e
m
po
r
a
r
y
bu
f
f
e
r
us
i
ng
t
e
m
pl
at
e
ba
s
e
d
a
p
pr
oa
c
h,
i
i
i
)
t
he
pr
o
p
os
e
d
s
y
s
t
em
do
e
s
n’
t
s
t
or
e
a
ny
r
a
w
da
t
a
o
ve
r
c
l
ou
d
bu
t
s
t
or
e
s
on
l
y
t
he
m
i
ne
d
da
t
a
w
hi
c
h
r
e
du
c
e
s
hi
gh
e
r
de
g
r
e
e
of
c
om
pl
e
xi
t
y
a
nd
i
nc
r
e
a
s
e
s
e
nr
i
c
hn
e
s
s
of
da
t
a
.
REFERE
NCE
S
[1]
L.
P
a
l
l
a
v
i
,
e
t
a
l
.,
“
E
R
MO
²
a
l
g
o
r
i
t
h
m
:
a
n
e
n
e
r
g
y
e
f
f
i
c
i
e
n
t
m
o
b
i
l
i
t
y
m
a
n
a
g
e
m
e
n
t
i
n
m
o
b
i
l
e
c
l
o
u
d
c
o
m
p
u
t
i
n
g
s
y
s
t
e
m
f
o
r
5
G
h
e
t
e
r
o
g
e
n
e
o
u
s
n
e
t
wo
r
k
s
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
C
o
m
pu
t
e
r
E
n
g
i
n
e
e
r
i
ng
(
I
J
E
C
E
)
,
v
o
l
.
9
,
n
o
.
3
,
p
p
.
1
9
57
-
1
9
6
7
,
2019
.
[2]
C
.
D
o
b
r
e
a
n
d
F
.
X
h
a
f
a
,
“
P
e
r
va
s
i
v
e
C
o
m
pu
t
i
n
g
:
N
e
x
t
G
e
n
e
r
a
t
i
o
n
P
l
a
t
f
o
r
m
s
fo
r
I
n
t
e
l
l
i
g
e
n
t
D
a
t
a
C
o
l
l
e
c
t
i
o
n
,
”
A
c
a
d
e
m
i
c
P
r
e
s
s
,
2
0
1
6
.
[3]
T.
F
r
a
n
c
i
s
a
n
d
M
.
M
a
d
h
i
a
j
a
g
a
n
,
“
A
C
om
p
a
r
i
s
on
o
f
C
l
o
u
d
E
x
e
c
u
t
i
o
n
M
e
c
h
a
n
i
sm
s
:
F
o
g
,
E
d
g
e
a
n
d
C
l
o
n
e
C
l
o
ud
C
o
m
pu
t
i
n
g
,
”
P
r
o
c
e
e
d
i
n
g
o
f
t
h
e
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
I
n
f
o
r
m
a
t
i
c
s
,
v
o
l
.
4
,
p
p
.
446
-
450
,
2017
.
[4]
F
.
M
.
G
r
o
om
a
n
d
S
.
S
.
J
o
n
e
s
,
“
En
t
e
r
p
r
i
s
e
C
l
o
u
d
C
o
m
pu
t
i
n
g
f
o
r
No
n
-
E
n
g
i
n
e
e
r
s
,
”
C
R
C
P
r
e
s
s
,
2
0
18
.
[5]
K
.
L
.
J
a
c
k
s
o
n
a
n
d
S
.
Go
e
s
s
l
i
ng
,
“
A
r
c
h
i
t
e
c
t
i
n
g
C
l
o
u
d
Co
m
p
u
tin
g
S
o
l
u
t
i
o
ns
:
B
u
i
l
d
c
l
o
u
d
s
t
ra
t
e
g
i
e
s
t
h
a
t
a
l
i
g
n
t
e
c
h
n
o
l
o
g
y
a
n
d
e
c
o
n
o
m
i
c
s
w
h
i
l
e
e
f
f
e
c
t
i
v
e
l
y
m
a
n
a
g
i
n
g
r
i
s
k
,
”
P
a
c
k
t
P
u
b
l
i
s
h
i
n
g
L
t
d
,
2
0
1
8
.
[6]
A
.
R
g
h
i
o
u
i
a
n
d
A
.
O
u
m
n
a
d
,
“
C
ha
l
l
e
n
g
e
s
a
n
d
o
p
p
o
r
t
u
n
i
t
i
e
s
o
f
i
n
t
e
r
n
e
t
o
f
t
h
i
n
g
s
i
n
h
e
a
l
t
h
c
a
r
e
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
na
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
ne
e
r
i
n
g
(
I
J
E
C
E)
,
v
o
l
.
8
,
n
o
.
5
, p
p
.
2753
-
2761
,
2018
.
[7]
B
.
R
i
s
t
e
v
s
k
i
a
n
d
M
.
C
h
e
n
,
“
B
i
g
D
a
t
a
A
n
a
l
y
t
i
c
s
i
n
M
e
d
i
c
i
n
e
a
n
d
H
e
a
l
t
h
c
a
r
e
,
”
J
o
u
r
n
a
l
o
f
I
n
t
e
gr
a
t
ive
B
i
o
i
n
f
o
r
m
a
t
i
c
s
,
vo
l
.
1
5
,
n
o
.
3
,
2
0
1
8
.
[8]
K
.
A
b
ou
e
l
m
e
h
di
,
e
t
a
l
.
,
“
B
i
g
h
e
a
l
t
h
c
a
r
e
d
a
t
a
:
p
r
e
s
e
r
v
i
n
g
s
e
cu
r
i
t
y
a
n
d
p
r
i
v
a
c
y,
”
J
o
u
r
n
a
l
o
f
B
i
g
D
a
t
a
,
vol
.
5
,
n
o.
1
,
2
0
1
8
.
[9]
Z
.
A
.
M
d
a
g
h
r
i
,
e
t
a
l
.
,
“
S
t
u
d
y
a
n
d
a
n
a
l
y
s
i
s
o
f
d
a
t
a
m
i
n
i
n
g
f
o
r
h
e
a
l
t
h
c
a
r
e
,
”
2
0
1
6
4
t
h
I
E
E
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
l
l
o
q
u
i
u
m
o
n
I
n
f
o
r
m
a
t
i
o
n
Sc
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
(
C
i
S
t)
,
T
a
n
g
i
e
r
,
p
p
.
7
7
-
82
,
2016
.
[
1
0]
Z
.
K
a
r
a
k
a
ya
,
“
So
f
t
w
a
r
e
e
n
g
i
n
e
e
r
i
n
g
i
s
s
u
es
i
n
b
i
g
d
a
t
a
a
p
p
l
i
c
a
t
i
o
n
d
e
v
e
l
o
p
m
e
n
t
,
”
2
0
1
7
I
n
t
e
r
n
a
t
i
o
na
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
(
U
B
MK)
,
A
n
t
a
l
y
a
,
p
p
.
8
5
1
-
855
,
2017
.
[
1
1]
A
.
M
u
s
l
im
,
e
t
a
l
.
,
“
W
e
b
s
e
r
v
ice
s
o
f
t
r
a
n
s
f
o
rm
a
t
i
o
n
d
a
t
a
b
a
s
e
d
o
n
O
p
e
n
E
HR
i
n
t
o
H
e
a
l
t
h
L
e
v
e
l
S
e
v
e
n
(
H
L
7
)
s
t
a
n
d
a
r
d
s
,
”
2
0
1
7
S
e
c
o
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
f
o
r
m
a
t
i
c
s
an
d
C
o
m
p
u
t
i
n
g
(
IC
I
C)
,
J
a
y
a
p
u
r
a
,
p
p
.
1
-
4
,
2017
.
[
1
2]
P
.
S
.
D
i
o
uf
,
e
t
a
l
.
,
“
V
a
r
i
e
t
y
o
f
d
a
t
a
i
n
t
h
e
E
T
L
p
r
o
c
e
s
s
e
s
i
n
t
h
e
c
l
o
u
d
:
S
t
a
t
e
o
f
t
h
e
a
r
t
,
”
2
0
1
8
I
E
E
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
n
n
o
v
a
t
i
v
e
R
e
s
e
a
r
c
h
a
n
d
D
e
v
e
l
o
p
m
e
n
t
(
I
C
I
RD)
,
B
a
n
g
k
o
k
,
p
p
.
1
-
5
,
2018
.
[
1
3]
Q
.
W
a
n
g
,
e
t
a
l
.
,
“
R
e
s
e
a
r
c
h
o
f
E
TL
o
n
u
n
i
v
e
r
s
i
t
y
d
a
t
a
e
x
c
h
a
n
g
e
p
l
a
t
f
o
r
m
,
”
2
0
1
1
IE
E
E
I
n
t
e
r
n
a
t
i
o
n
a
l
S
y
m
p
o
s
i
um
on
I
T
i
n
M
e
d
i
c
i
n
e
a
n
d
E
d
u
c
a
t
i
o
n
,
C
u
a
n
g
z
h
o
u
,
p
p
.
2
8
5
-
288
,
2011
.
[
1
4]
Y
.
W
a
n
g
,
e
t
a
l
.,
“
B
i
g
d
a
t
a
a
n
a
l
y
t
i
c
s
:
U
n
d
e
r
s
t
a
n
d
i
n
g
i
t
s
c
a
p
a
b
i
l
i
t
i
e
s
a
n
d
p
o
t
e
n
t
i
a
l
b
e
n
e
f
i
t
s
f
o
r
h
e
a
l
t
h
c
a
r
e
o
r
g
a
n
i
z
a
t
i
o
n
s
,”
T
e
c
h
n
o
l
o
g
i
c
a
l
F
o
r
e
c
a
s
t
i
n
g
a
n
d
S
oc
i
a
l
C
h
a
n
g
e
,
v
o
l
.
126
,
p
p
.
3
-
13
,
2
0
1
8
.
[
1
5]
J
.
A
.
P
e
r
e
z
,
e
t
a
l
.
,
“
B
i
g
D
a
t
a
f
o
r
H
e
a
l
t
h
,
”
in
I
E
E
E
J
o
u
r
n
a
l
o
f
B
i
o
m
e
d
i
c
a
l
a
n
d
He
a
l
t
h
I
n
f
o
r
m
a
t
i
cs
,
v
o
l
.
1
9
,
n
o
.
4
,
p
p
.
1
1
9
3
-
1
2
0
8
,
Ju
l
.
2015
.
[
1
6]
S
.
T
.
P
r
a
s
a
d
,
e
t
al
.
,
“
D
i
a
b
e
t
i
c
d
a
t
a
a
n
a
l
ys
i
s
i
n
b
i
g
d
a
t
a
w
i
t
h
p
r
e
d
i
c
t
i
v
e
m
e
t
h
o
d
,
”
2
0
1
7
I
n
t
e
r
n
a
t
i
o
na
l
C
o
n
f
e
r
e
n
c
e
o
n
Alg
o
r
i
t
h
m
s
,
M
e
t
h
o
d
o
l
o
g
y
,
M
o
d
e
l
s
&
Ap
p
l
i
c
a
t
i
o
n
s
i
n
E
m
e
r
g
i
n
g
T
e
c
h
n
o
l
o
g
i
e
s
(
IC
A
M
M
A
E
T)
,
C
h
e
n
n
a
i
,
p
p
.
1
-
4
,
2017
.
[
1
7]
N
.
D
a
s
,
e
t
a
l
.,
“
B
i
g
D
a
t
a
A
n
a
lyti
c
s
f
o
r
M
e
d
i
c
a
l
A
p
p
l
i
c
a
t
i
o
n
s
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
M
o
de
r
n
E
d
u
c
a
t
i
o
n
an
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
10
,
n
o
.
2,
p
p.
35
,
2018
.
[
1
8]
I
.
G
.
M
a
g
a
r
i
ñ
o
,
e
t
a
l
.
,
“
A
g
e
n
t
-
B
a
s
e
d
S
im
u
l
a
t
i
on
o
f
S
m
a
r
t
B
e
d
s
W
i
t
h
I
n
t
e
r
n
e
t
-
of
-
T
h
i
n
g
s
fo
r
Ex
p
l
o
r
i
n
g
B
i
g
D
ata
A
n
a
l
y
t
i
c
s
,
”
in
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
6
,
p
p
.
366
-
3
7
9
,
2
0
1
8
.
[
1
9]
H
.
Z
h
o
n
g
a
n
d
J
.
X
i
a
o
,
“
E
n
h
a
n
c
i
n
g
h
e
a
l
t
h
r
i
s
k
p
re
d
i
c
t
i
o
n
w
i
t
h
d
e
e
p
l
e
a
r
n
i
n
g
o
n
bi
g
d
a
t
a
a
n
d
r
e
v
i
s
e
d
f
u
s
i
on
n
o
d
e
p
a
r
a
d
i
g
m
,”
S
c
i
e
n
t
i
f
i
c
P
r
o
g
r
a
m
m
in
g
,
2017
.
[
2
0]
M
.
C
h
e
n
,
e
t
a
l
.
,
“
D
is
e
a
s
e
P
r
e
d
i
c
t
i
o
n
b
y
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
O
v
e
r
B
i
g
D
a
t
a
F
r
om
He
a
l
t
h
c
a
r
e
C
o
m
m
u
n
i
t
i
e
s
,
”
in
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
5
,
p
p
.
8
8
6
9
-
8
8
7
9
,
2
01
7
.
[
2
1]
Y
.
Z
h
a
n
g
,
e
t
a
l
.
,
“
H
e
a
l
t
h
-
C
PS
:
H
e
a
l
t
h
c
a
r
e
C
yb
e
r
-
P
h
y
s
i
c
a
l
S
ys
t
e
m
A
ss
i
s
t
e
d
by
C
l
o
u
d
a
n
d
B
ig
D
a
t
a
,
”
in
I
E
E
E
S
y
s
t
e
m
s
J
o
u
r
na
l
,
v
o
l
.
1
1
,
n
o
.
1
,
p
p
.
8
8
-
9
5
,
M
a
r
.
2
0
1
7
.
[
2
2]
P
.
W
u
,
e
t
a
l
.
,
“
Om
i
c
a
n
d
E
l
ect
r
o
n
i
c
H
e
a
l
t
h
R
e
c
o
r
d
B
i
g
D
ata
A
n
a
l
y
t
i
c
s
f
o
r
P
r
e
c
i
s
i
o
n
M
e
di
c
i
n
e
,
”
in
I
E
E
E
T
r
a
ns
a
c
t
i
o
n
s
o
n
B
i
o
m
e
d
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
6
4
,
n
o
.
2
,
p
p
.
2
6
3
-
2
7
3
,
F
e
b
.
2017.
[
2
3]
M
.
V
i
c
e
c
o
n
t
i
,
e
t
a
l
.
,
“
B
i
g
D
a
t
a
,
B
i
g
K
n
o
w
l
e
d
g
e
:
B
i
g
D
a
t
a
f
o
r
P
e
r
s
o
n
a
l
i
z
e
d
H
e
a
l
t
h
c
a
r
e
,
”
in
I
E
E
E
J
o
u
r
n
a
l
of
B
i
o
m
e
d
i
c
a
l
a
n
d
H
e
a
l
t
h
I
n
f
o
r
m
a
t
ics
,
v
o
l
.
1
9
,
n
o
.
4
,
p
p
.
1
2
0
9
-
1
2
1
5
,
J
u
l
.
2015
.
[
2
4]
S
.
S
h
a
f
q
a
t
,
e
t
a
l
.,
“
B
i
g
d
a
t
a
a
n
a
l
y
t
i
c
s
e
n
h
a
n
c
e
d
h
e
a
l
t
h
c
a
r
e
s
y
s
t
e
m
s
:
a
r
e
v
i
e
w
,”
T
h
e
J
o
u
r
n
a
l
o
f
S
u
p
e
r
c
o
m
p
u
t
i
n
g
,
p
p
.
1
-
46
,
2018
.
[
2
5]
M
.
C
h
e
n
,
e
t
a
l
.
,
“
5G
-
Sm
a
r
t
D
i
a
be
t
e
s
:
T
o
w
a
r
d
P
e
r
s
o
n
a
l
i
z
e
d
D
i
a
b
e
t
e
s
D
i
a
g
n
o
s
i
s
w
ith
H
e
a
l
t
h
c
a
r
e
B
i
g
D
a
t
a
C
l
o
u
d
s
,
”
in
I
E
E
E
C
o
m
m
u
n
i
c
a
t
i
o
n
s
M
a
g
a
zi
n
e
,
v
o
l
.
5
6
,
n
o
.
4
,
p
p
.
1
6
-
2
3
,
A
pr
.
2018.
[
2
6]
C.
G
a
r
a
t
t
i
n
i
,
e
t
a
l
.,
“
B
i
g
D
a
t
a
A
n
a
l
yt
i
c
s
,
I
n
f
e
c
t
i
o
u
s
D
i
s
e
a
s
es
a
n
d
A
ss
o
c
i
a
t
e
d
E
t
h
i
c
a
l
I
m
p
a
c
t
s
.”
P
h
i
l
o
s
o
p
hy
&
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
3
2
,
p
p
.
6
9
-
85
,
2
0
1
7
.
[
2
7]
D
.
C
h
r
i
m
e
s
a
n
d
H
.
Z
a
m
a
n
i
,
“
U
s
i
n
g
D
i
s
t
r
i
b
u
t
e
d
D
a
t
a
o
v
e
r
H
B
a
s
e
i
n
B
i
g
D
a
t
a
A
n
a
l
y
t
i
c
s
P
l
a
t
f
o
r
m
f
o
r
C
l
i
n
i
c
a
l
S
e
r
v
i
c
e
s
,”
C
o
m
pu
t
a
t
i
o
n
a
l
a
n
d
M
a
t
h
e
m
a
t
i
c
a
l
M
e
t
h
o
d
s
i
n
M
e
d
i
c
i
n
e
,
v
o
l
.
2
0
1
7
,
p
p
.
1
-
1
6
,
2017
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
48
53
-
48
62
4862
[
2
8]
L
.
A
.
T
a
w
a
l
b
e
h
,
e
t
a
l
.
,
“
M
o
b
i
l
e
C
l
o
u
d
C
o
m
p
u
t
in
g
M
o
d
e
l
a
n
d
B
i
g
D
a
t
a
A
n
a
l
ys
i
s
f
o
r
H
e
a
l
t
h
c
a
r
e
A
p
p
l
i
c
a
t
i
o
n
s
,
”
in
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
4
,
p
p
.
6
1
7
1
-
6
1
8
0
,
2
0
1
6
.
[
2
9]
B
.
K
.
S
a
r
k
a
r
,
“
Bi
g
d
a
t
a
f
o
r
s
e
c
u
r
e
h
e
a
l
t
h
c
a
r
e
s
y
s
t
e
m
:
a
c
o
n
c
e
p
t
u
a
l
d
e
s
i
g
n
,”
C
o
m
p
l
e
x
&
In
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
,
v
o
l
.
3
,
n
o
.
2
,
p
p
.
133
-
151
,
2017
.
[
3
0]
Y
.
Z
h
a
n
g
,
e
t
a
l
.
,
“
I
EEE
A
c
c
e
s
s
S
p
e
c
i
a
l
S
e
c
t
i
o
n
E
d
i
t
o
r
i
a
l
:
B
i
g
D
a
t
a
A
n
a
l
yt
i
c
s
f
o
r
S
m
a
r
t
a
n
d
C
o
n
n
e
c
t
e
d
H
e
a
l
t
h
,
”
in
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
4
,
p
p
.
9
9
0
6
-
9
9
0
9
,
2
0
1
6
.
[
3
1]
U
.
S
r
i
n
i
v
as
a
n
a
nd
B
.
Ar
u
n
a
s
a
l
a
m
,
“
L
e
v
e
r
a
g
i
n
g
B
ig
D
a
t
a
An
a
l
y
t
i
c
s
t
o
R
e
d
u
c
e
H
e
a
l
t
h
c
a
r
e
C
o
s
t
s
,
”
in
I
T
P
ro
f
e
s
s
i
o
na
l
,
v
o
l
.
1
5
,
n
o
.
6
,
p
p
.
2
1
-
2
8
,
2
0
1
3
.
[
3
2]
R
.
B
h
a
l
l
a
a
n
d
A
.
B
a
g
g
a
,“
O
p
i
ni
o
n
m
i
n
i
n
g
f
r
am
e
w
o
rk
u
s
i
ng
p
r
o
p
o
s
e
d
r
b
-
b
a
ye
s
m
o
d
e
l
f
o
r
t
ex
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
(
I
JEC
E
)
,
v
o
l
.
9
,
n
o
.
1
,
p
p
.
4
7
7
-
4
8
5
,
2019
.
[
3
3]
S
.
A
i
c
h
,
e
t
a
l
.,
“
C
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
-
b
a
s
e
d
m
o
d
e
l
f
o
r
we
b
-
b
a
s
e
d
t
e
x
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
al
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
ne
e
r
i
n
g
(
I
J
E
C
E)
,
v
o
l
.
9
,
n
o
.
6
,
p
p
.
5
1
8
5
-
51
9
1
,
2019
.
[
3
4]
“
C
e
n
su
s
D
a
t
a
,
”
U
n
i
t
e
d
S
t
a
t
e
C
e
n
s
u
s
,
[
O
n
l
i
n
e]
.
A
va
i
l
a
b
l
e
:
h
t
t
p
:
/
/
w
w
w
.
c
e
n
s
us
.
g
o
v
/
da
t
a
.
h
t
m
l
.
[
R
e
t
r
i
e
v
e
d
11
-
04
-
2020
].
[
3
5]
“
P
h
y
s
i
o
n
e
t
,
”
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
w
w
w
.
p
h
y
s
i
o
n
e
t
.
o
r
g
/
p
h
y
s
i
o
b
a
n
k
/
d
a
t
a
b
a
s
e
/
.
[
R
e
t
r
i
e
v
e
d
o
n
1
1
-
04
-
2020
]
[
3
6]
“
R
e
g
i
s
t
r
y
o
f
O
p
e
n
D
a
t
a
o
n
A
W
S
,
”
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
r
e
g
i
s
t
r
y
.
o
p
e
n
d
a
t
a
.
a
w
s
/
,
[
R
e
t
r
i
e
v
e
d
o
n
1
1
-
04
-
2
0
20
].
[
3
7]
“
NH
S
C
o
n
t
i
n
u
i
ng
H
e
a
l
t
h
c
a
r
e
A
c
t
i
v
i
t
y
,
”
d
a
t
a
.
g
o
v
.
u
k
.
[
O
n
l
i
n
e]
.
A
va
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
d
a
t
a
.
g
o
v
.
u
k
/
d
a
t
a
s
e
t
/
f
2
5
9
a
7
b
7
-
6
c
9
7
-
4632
-
93d0
-
3
8
b
93
a
f
b
3
2
4
f
/
n
h
s
-
c
on
t
i
n
u
i
n
g
-
h
e
a
l
t
h
c
a
r
e
-
a
c
t
i
v
i
t
y
,
R
e
t
r
i
e
v
e
d
o
n
1
1
-
04
-
2020
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Chandr
u
A
S
,
Assistant
P
rof
essor,
De
pa
rtm
ent
of
I
nfor
m
ation
Scie
nce
&
En
gin
eeri
ng
,
NI
E
Insti
tute
of
Tec
hnol
og
y
,
My
su
r
u,
I
nd
ia
.
He
is
purs
uing
his
P
hD
from
VTU
,
Be
la
gav
i,
I
nd
i
a.
His
area
of
interest
is
Bi
g
data
a
naly
tics.
He
has
8
ye
ars
of
w
ork
exp
e
rience
.
S
e
e
t
h
ar
am
K
e
s
h
av
ar
a
o,
Pro
f
essor,
Dep
a
rtm
ent
of
Com
pu
t
er
Scie
nce
&
En
gin
eeri
ng
,
Jn
a
nav
i
kas
In
s
ti
tute
of
Tech
no
l
og
y,
Bi
da
di,
Be
ngal
uru
,
I
nd
ia
.
He
ha
s
c
om
plete
d
his
PhD
f
ro
m
II
T
Bom
bay.
His
area
of
interest
was
arti
fici
al
intel
li
gen
ce.
H
e
has
ar
ound
34 yea
rs o
f work
experie
nce.
Evaluation Warning : The document was created with Spire.PDF for Python.