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.
15
,
No.
1
,
Febr
uary
20
25
, pp.
1
175
~
1
186
IS
S
N:
20
88
-
8708
, DO
I:
10
.11
591/ij
ece.v
15
i
1
.
pp
1
175
-
1
186
1175
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om
Deep
l
ea
rn
i
ng fo
r
i
nfectious
d
iseas
e
s
urveilla
nce
i
ntegr
ating
internet
o
f thi
ngs
for
r
ap
i
d
r
esp
on
se
Subr
amanian
Sumithr
a
1
,
M
oo
r
th
y
Radhi
ka
2
,
G
an
d
avadi
Ven
katesh
3
,
Bab
u
Se
eth
a L
ak
shmi
4
,
Balraj
Vic
to
ri
a
Jancee
5
,
N
agar
aja
n
Mohankum
a
r
6
,
Sub
biah
Mur
uga
n
7
1
Dep
artm
en
t of
E
l
ectron
ics an
d
Co
m
m
u
n
icatio
n
E
n
g
in
eering
,
J.J. Co
lleg
e
o
f
Eng
in
eering
and
T
echn
o
lo
g
y
,
Tir
u
ch
irapp
alli,
Ind
ia
2
Dep
artm
en
t of
I
n
f
o
rm
atio
n
T
echn
o
lo
g
y
,
R.M.
D
E
n
g
in
eering
Co
lleg
e,
Ch
en
n
ai,
Ind
ia
3
Dep
artm
en
t of
Co
m
p
u
ter
Sci
en
ce a
n
d
E
n
g
in
eering
,
Ko
n
eru La
k
sh
m
aiah
Edu
catio
n
Fou
n
d
at
io
n
,
Vad
d
eswara
m
,
Ind
ia
4
Dep
artm
en
t of
Ar
tificial
Intellig
en
ce
and
Data
S
cien
ce,
Sr
i Shan
m
u
g
h
a Co
lleg
e of E
n
g
in
eeri
n
g
and
T
echn
o
lo
g
y
,
Sale
m
,
Ind
ia
5
Dep
artm
en
t of
E
l
ectron
ics
an
d
Co
m
m
u
n
icatio
n
E
n
g
in
eering
,
St.
Jo
sep
h
’s
Co
lleg
e of E
n
g
in
eering
,
Ch
en
n
ai,
In
d
ia
6
Dep
artm
en
t
o
f
Co
m
p
u
ter
Sci
en
ce a
n
d
E
n
g
in
eering
,
Sy
m
b
io
sis
I
n
stitu
te of T
e
ch
n
o
lo
g
y
,
Nag
p
u
r
Campu
s, Sym
b
io
sis
I
n
ternatio
n
al
(Dee
m
ed
Univ
ersit
y
),
Pun
e,
Ind
ia
7
Dep
artm
en
t of
B
i
o
m
ed
ical E
n
g
in
eer
in
g
,
Sav
eeth
a Scho
o
l of E
n
g
in
eering
,
Sav
eeth
a I
n
stitu
te
o
f
Medical
and
T
e
ch
n
ical Sciences,
Sav
eeth
a Univ
ersit
y
,
Ch
en
n
ai,
Ind
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
J
un
29, 2
024
Re
vised
Sep 1
0,
2024
Accepte
d
Oct
1,
2024
Parti
cularl
y
in
t
he
c
ase
of
eme
rging
inf
ec
t
ious
disea
ses
and
worldwide
pande
m
ic
s,
inf
ectious
dise
ase
mo
nit
oring
is
essenti
al
for
qui
ck
identifi
c
ation
and
eff
ic
i
ent
r
e
spons
e
to
epi
d
e
mi
cs.
Improvi
n
g
surveil
l
ance
sys
te
ms
for
quic
k
react
ion
mi
ght
be
poss
ibl
e
w
i
th
the
he
lp
of
n
ew
d
ee
p
l
e
arn
ing
and
int
ern
et
of
thi
n
gs
(IoT
)
t
ec
hno
logi
es.
Th
is
p
ap
er
int
roduc
es
an
inf
ec
t
ious
disea
se
mon
it
ori
ng
arc
hi
tectur
e
base
d
on
de
ep
l
ea
rning
coupled
with
IoT
devi
c
es
to
f
acilitate
e
arl
y
dia
gn
osis
and
proa
ct
i
ve
interve
n
ti
on
me
asur
es
.
Thi
s
appr
o
ac
h
u
ses
rec
urre
n
t
ne
ura
l
n
et
works
(
RNN
s)
to
ide
ntify
te
mpor
al
pat
t
ern
s
suggesti
ve
of
infect
ious
disea
se
outbr
eaks
by
an
al
yz
ing
seque
nt
ia
l
dat
a
ret
r
ie
v
ed
f
rom
IoT
device
s
li
ke
smart
th
erm
om
et
ers
and
wea
rab
le
sensors
.
To
id
e
nti
fy
sm
all
changes
in
he
al
th
ma
rke
rs
and
fo
rec
ast
th
e
deve
lop
me
nt
of
disea
ses,
RNN
arc
hi
te
c
ture
s
wit
h
long
short
-
te
r
m
me
mo
r
y
(LSTM)
net
wor
ks
are
used
to
c
apt
ure
long
-
r
ang
e
r
el
a
ti
onships
i
n
the
da
ta.
Spati
al
an
al
ysis
per
mi
ts
the
in
te
g
rat
ion
of
g
eogr
a
phic
d
ata
fro
m
I
oT
dev
ice
s
,
al
lowing
for
th
e
id
ent
if
ic
a
ti
on
of
infect
ion
hot
spots
and
th
e
t
rac
king
of
aff
licte
d
per
son
s'
move
me
n
ts.
Quick
a
ct
ion
st
eps
li
ke
foc
use
d
te
sting
,
cont
a
ct
tracin
g
,
and
me
d
ical
resourc
e
d
eployment
ar
e
pr
ompt
ed
by
abnor
malities
d
et
e
ct
ed
early
by
real
-
time
m
onit
oring
and
ana
lysis
.
Preve
nti
ng
or
l
e
ss
eni
ng
the
sev
eri
ty
of
inf
ectio
us
disea
se
ou
tbr
ea
ks
is
the
goal
of
the
pla
n
ned
monitoring
sys
te
m,
whi
ch
would
enha
n
ce
publi
c
h
ealt
h
rea
din
ess a
nd
r
e
spons
e
ca
pa
cities
.
Ke
yw
or
d
s
:
Data inte
grat
io
n
Healt
h
m
onit
ori
ng
Pu
blic healt
h
Re
al
-
ti
me mon
it
or
in
g
Seco
nd k
e
ywo
rd
Time
ser
ie
s
an
al
ys
is
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
BY
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Nag
a
rajan
Mo
hank
um
a
r
Dep
a
rtme
nt of
Com
pu
te
r
Scie
nce a
nd E
ng
i
ne
erin
g,
Symbi
osi
s I
nst
it
ute of
Tech
no
l
ogy, N
agpu
r
Ca
m
pus,
Sy
m
b
io
sis I
nte
rn
at
io
nal (Dee
med U
niv
e
rsity)
Pune, I
nd
ia
Emai
l:
nmkpr
ofesso
r
@gmai
l.com
1.
INTROD
U
CTION
T
o
a
d
d
r
e
s
s
t
h
e
p
r
o
b
l
e
m
o
f
m
a
n
y
c
o
r
o
n
a
v
i
r
u
s
d
i
s
e
a
s
e
2019
(
C
O
V
I
D
-
19
)
p
a
t
i
e
n
t
s
w
i
t
h
p
a
pe
r
f
o
r
m
s
f
o
r
i
n
v
e
s
t
i
g
a
t
i
o
n
a
n
d
m
e
d
i
c
a
l
s
t
a
f
f
-
p
u
b
l
i
c
h
e
a
l
t
h
u
n
i
t
c
o
m
m
un
i
c
a
t
i
o
n
.
W
e
de
s
i
g
n
e
d
a
n
d
d
e
v
e
l
o
p
e
d
a
n
i
n
f
e
c
t
i
o
u
s
d
i
s
e
a
s
e
m
o
n
i
t
o
r
i
n
g
s
y
s
t
e
m
d
a
t
a
t
r
a
n
s
l
a
t
i
o
n
us
i
n
g
t
h
e
a
g
i
l
e
m
e
t
h
o
d
o
l
o
g
y
t
o
a
l
l
o
w
a
p
r
o
j
e
c
t
t
o
r
e
a
c
t
t
o
c
h
a
n
g
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
1
175
-
1
186
1176
r
e
q
u
e
s
t
s
q
u
i
c
kl
y
[1]
.
T
h
e
s
e
d
a
t
a
m
o
d
i
f
i
c
a
t
i
o
n
s
e
n
a
b
l
e
r
e
g
u
l
a
r
,
C
O
V
I
D
-
1
9
,
e
v
e
n
t
-
b
a
s
e
d
,
a
n
d
e
p
i
d
e
m
i
c
s
u
r
v
e
i
l
l
a
n
c
e
i
n
p
u
b
l
i
c
h
e
a
l
t
h
e
m
e
r
g
e
n
c
i
e
s
.
E
m
e
r
g
i
n
g
i
n
f
e
c
t
i
o
u
s
d
i
s
e
a
s
e
p
r
e
v
e
n
t
i
o
n
,
c
o
n
t
a
i
n
m
e
n
t
,
a
n
d
p
r
e
d
i
c
t
i
o
n
a
r
e
e
s
s
e
n
t
i
a
l
f
o
r
m
a
n
a
g
i
n
g
t
h
e
i
r
s
p
r
e
a
d
a
n
d
i
m
p
a
c
t
[2]
.
I
n
f
e
c
t
i
o
u
s
i
l
l
n
e
s
s
m
a
na
g
e
m
e
n
t
h
a
s
u
s
e
d
s
e
v
e
r
a
l
m
o
d
e
l
i
n
g
m
e
t
h
o
d
s
.
E
m
e
r
g
i
n
g
i
l
l
n
e
s
s
e
s
a
r
e
c
o
n
n
e
c
t
e
d
t
o
s
e
l
e
c
t
e
d
h
u
m
a
n
f
o
r
c
e
s
l
i
k
e
u
r
b
a
n
i
z
a
t
i
o
n
a
n
d
h
a
b
i
ta
t
f
r
a
g
m
e
n
t
a
t
i
o
n
.
I
n
f
e
c
t
i
o
u
s
i
l
l
n
e
s
s
p
a
n
d
e
m
i
c
s
d
a
m
a
g
e
l
i
v
e
s
a
n
d
p
r
o
p
e
r
t
y
a
n
d
p
r
o
v
o
k
e
f
e
a
r
a
n
d
u
n
r
e
s
t
.
E
s
t
a
bl
i
s
hi
n
g
a
n
i
n
f
e
c
t
i
o
u
s
d
i
s
e
a
s
e
e
a
r
l
y
w
a
r
n
i
n
g
p
l
a
t
f
o
r
m
a
n
d
t
a
k
i
n
g
t
i
m
e
l
y
a
c
t
i
o
n
s
t
o
a
v
e
r
t
o
u
t
b
r
e
a
k
s
i
s
c
r
u
c
i
a
l
[3]
.
I
t
i
n
c
l
u
d
e
s
i
n
f
e
c
t
i
o
u
s
d
i
s
e
a
s
e
e
a
r
l
y
w
a
r
n
i
ng
s
y
s
t
e
m
t
e
r
m
s
,
n
a
t
i
o
n
a
l
a
n
d
g
l
o
b
a
l
d
e
v
e
l
o
p
m
e
n
t
s
t
a
t
u
s
,
i
n
d
i
c
a
t
o
r
s
o
f
i
n
f
e
c
t
i
o
n
a
n
d
s
y
s
t
e
m
s
,
a
n
d
s
o
u
r
c
e
s
f
o
r
e
s
t
a
b
l
i
s
h
i
n
g
t
h
e
m
.
R
e
a
l
-
w
o
r
l
d
a
p
p
l
i
c
a
t
i
o
ns
o
f
t
e
n
f
a
c
e
s
i
m
i
l
a
r
i
m
p
u
t
a
t
i
o
n
c
h
a
l
l
e
n
g
e
s
t
h
a
t
a
r
e
d
i
f
f
i
c
u
l
t
t
o
s
o
l
v
e
[4]
.
K
e
r
n
e
l
-
b
a
s
e
d
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
u
s
e
s
l
o
c
a
t
i
o
n
-
s
p
e
c
i
f
i
c
d
i
s
e
a
s
e
-
r
e
l
a
t
e
d
r
i
s
k
f
a
c
t
o
r
f
e
a
t
u
r
e
s
f
r
o
m
d
i
v
e
r
s
e
d
a
t
a
s
o
u
r
c
e
s
t
o
a
d
d
r
e
s
s
t
h
i
s
d
i
f
f
i
c
u
l
t
c
h
a
l
l
e
n
g
e
.
Pr
e
ven
ti
on,
he
al
th,
a
nd
s
ocial
go
od
de
pe
nd
on
earl
y
detect
ion
an
d
c
ontr
ol
of
i
nf
ect
io
us
il
lnesses
vi
a
act
ive
m
onit
or
i
ng
[5]
.
I
ntell
igently
pic
king
a
small
gro
up
of
no
des
as
senti
nels
from
ma
ny
pe
rsons
to
id
entify
infecti
ous
dise
ase
outb
reaks
early
is
a
t
ou
gh
ta
sk
i
n
act
ive
sur
veill
ance
[6]
.
Existi
ng
se
ntinel
sa
mp
li
ng
te
chn
iq
ues
ba
s
ed
on
global
s
ocial
net
work
inf
or
mati
on
co
uld
be
m
or
e
e
ff
ic
ie
nt
a
nd
preci
se.
S
om
e
stud
ie
s
heurist
ic
al
ly
pi
ck
se
ntinels
usi
ng
l
ocal
kn
owle
dg
e
a
bout
li
nked
nei
ghbors
.
Few
co
ns
id
er
the
ti
me
str
uct
ur
e
of
so
ci
al
ti
es,
whic
h
ma
y
direct
ly
af
fect
in
fec
ti
ou
s
il
lness
pro
pa
gation
[
7]
.
Ba
sed
on
t
he
fr
ie
ndsh
i
p
pa
rado
x
hypothesis,
w
hic
h
sta
te
s
th
at
mo
st
in
divi
du
al
s
hav
e
f
ewer
fr
ie
nds
than
t
heir
f
ri
ends,
we
offe
r
tw
o
te
mp
oral
-
netw
ork
s
urveil
la
nc
e
al
gorithms
f
or
c
hoosi
ng
se
ntinels.
Infecti
ou
s
m
on
it
or
in
g
is
on
e
of
t
he
mo
st
com
plete
s
ys
te
ms
for
colle
ct
ing,
a
naly
zi
ng,
and
inte
rpreti
ng
data
on
infe
c
ti
ou
s
disease
prolife
rati
on,
ve
ct
or
s,
and
ou
t
br
ea
ks
.
Additi
on
al
l
y,
da
ta
sho
uld
be
pro
vid
e
d
quic
kl
y
t
o
regulat
e
i
nfect
iou
s
il
lness
pre
ven
ti
on
[
8]
.
An
infecti
ous
il
lne
ss
pre
dicti
on
model
is
c
ru
ci
al
for
real
-
ti
m
e
ma
nag
e
ment
.
Co
ns
tr
uct
a
model
util
iz
ing
l
ong
sh
ort
-
te
r
m
me
mory
(L
ST
M
)
that
can
f
oreca
st
the
e
pid
e
mic
disease's
locat
ion
an
d
i
ntensi
ty.
Infecti
ous
di
sease
su
r
veill
ance is
an
esse
ntial
as
pect of
healt
hc
are,
as
s
how
n
i
n
Fi
gure
1.
Figure
1. I
nf
ec
ti
ou
s
disease s
urveil
la
nce
hea
lt
hcar
e esse
ntial
s
This
work
pr
ovides
a
cr
owd
s
ourcin
g
te
c
hniqu
e
for
l
ow
-
c
os
t,
real
-
ti
me
commu
nity
m
on
it
ori
ng
a
nd
mass
te
m
per
at
ur
e
s
cree
ning
modu
le
s
base
d
on
t
hermal
imagin
g
[9]
.
This
is
in
res
pons
e
t
o
the
curren
t
epidemic
of
in
fecti
ou
s
il
lness
es
in
huma
ns
a
nd
animal
s
.
T
he
refor
e
,
plan
nin
g
an
d
e
xec
uting
th
oro
ugh
ac
ti
ons
to
a
void
disea
ses,
ma
nag
e
th
em
e
ff
ect
ivel
y,
a
nd
f
oreca
st
wh
e
n
the
y
wil
l
sprea
d
is
c
riti
cal
[10
]
.
Yet,
the
r
e
needs
to
be
an
appr
opriat
e
co
mp
a
rison
of
ti
me
ser
ie
s
a
nalytic
meth
od
s
f
or
var
i
ous
il
lnesses,
s
uch
a
s
measl
es
,
deng
ue
fe
ver,
and
ha
nd
-
f
oot
-
mouth
[
11]
.
Time
series
anal
ys
is
will
be
us
ed
in
this
re
se
arch
to
c
ompa
re
the
Seaso
nal
ARI
M
A
m
odel
w
it
h othe
r
m
odel
s for
f
or
ecast
in
g t
he
hand f
oo
t
mouth,
d
e
ngue
, and mea
sle
s
.
Infecti
ous
il
lnesses
are
to
p
global
healt
h
con
ce
r
ns
.
D
ue
to
human
,
bio
lo
gical
,
c
li
mati
c,
and
ecolo
gical
cau
ses,
ma
ny
il
lnesses
emer
ge
d.
Infecti
ous
in
fe
ct
ion
s
hav
e
spr
ead
w
orl
dwide
in
the
recent
de
cade
[12]
.
T
his
co
ndit
ion
em
phasi
zes
the
neces
sit
y
for
a
ra
pi
d
disease
detect
i
on
s
ys
te
m
t
o
de
te
ct
,
diag
no
se
,
an
d
con
t
ro
l
pa
nde
mic
diseases.
T
he
I
nd
ia
n
su
r
veill
ance
s
ys
te
m,
par
ti
cu
la
rly
Ke
rala,
is
re
viewe
d.
Existi
ng
su
r
veill
ance
s
yst
ems
ca
n
detect
global
or
national
epi
de
mics
[
13]
.
It
c
an
no
ti
f
y
healt
h
prof
e
ssio
nal
s
with
detai
le
d
data.
Inf
ect
io
us
il
lnes
ses lea
d
t
o
m
o
r
ta
li
ty an
d
ec
on
om
ic
dama
ge g
lob
al
ly.
A m
ore resil
ie
nt,
a
da
ptive,
and
a
da
ptable
structu
re
w
ould
e
nhan
ce
e
pide
mic
re
spo
ns
e
[14]
.
En
gin
ee
r
s
bu
il
d
a
nd
c
onstr
uct
i
nfrastr
uctu
r
e
so
novel
in
fecti
ou
s
diseas
e
tr
eat
ment
te
c
hnol
o
gies
ma
y
be
impleme
nted
in
c
urre
nt
str
uc
tures
with
ef
fi
ci
enc
y
and
res
ources
befor
e
a
public
he
al
th
disaste
r
.
Ou
t
br
ea
ks
m
ay
over
w
helm
fr
ai
l
healt
h
s
yst
ems
that
need
m
or
e
resou
rces,
i
nfr
ast
ru
ct
ure,
re
gula
ti
on
s
, and p
ro
t
oco
ls t
o protec
t pop
ulati
on
s
. P
a
kistan
needs a centr
al
iz
ed heal
th
and
disease
s
urveil
la
nce
sy
ste
m.
P
hy
sic
al
a
nd
el
ect
ronic
m
edia
re
port
ai
lments
in
t
he
pr
esent
ma
nu
al
di
sease
monit
or
i
ng
s
yst
em.
Du
e
to
th
e
dela
y,
epi
de
mics
are
re
port
ed
i
n
pr
i
nt
a
nd
el
ect
ronic
m
e
dia
[
15]
.
T
he
s
ys
te
m
disease
outb
re
ak
co
unte
rmea
su
res
an
d
ma
y
in
du
ce
wi
de
sp
rea
d
pan
ic
.
It
pro
vid
es
a
cl
oud
-
base
d
healt
h
mana
geme
nt
s
ys
te
m
with
tre
nd
anal
ys
is
for
il
lness
m
on
it
ori
ng
a
nd
early
warnin
g.
P
ub
li
c
healt
h
officer
s
ma
y
Inf
ectio
u
s Diseas
e
Su
rveillan
ce
in
Health
care
Patien
t
Health
Reco
rds
(dem
o
g
raph
ics,
sym
p
to
m
s
,
d
iag
n
o
ses
,
treatm
en
ts,
etc.)
Clin
ical
Data
(Sy
m
p
to
m
s,
Diag
n
o
ses
,
Tr
eatm
en
t,
Ou
tco
m
e,
etc.)
Epid
em
io
lo
g
ical
Su
rveillan
ce
(ou
tb
reak,
case
repo
rting
,
d
iseas
e m
ap
p
in
g
,
co
n
tact tr
acin
g
,
etc.)
Labo
ratory
Tests
(m
icrob
io
lo
g
ical,
sero
lo
g
ical,
an
d
m
o
lecu
lar
testin
g
)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
omp E
ng
IS
S
N:
20
88
-
8708
Deep l
ear
ning
for infect
io
us
di
sease
s
ur
vei
ll
an
ce
integ
ra
ti
ng
inter
net
of th
ing
s
…
(
Subra
manian
Sumith
ra
)
1177
employ
ti
mely
and
meani
ngf
ul
healt
h
d
at
a
in
a
su
r
veill
ance
sy
ste
m
a
nd
a
da
ta
-
dr
i
ven
f
or
e
cast
ing
m
odel
[16]
.
The
disease
s
urveil
la
nce
s
ys
t
em
ma
y
detect
public
healt
h
hazar
ds
a
nd
m
ulti
-
disease
outbrea
ks
.
T
he
major
pur
po
se
of
t
his
co
mputer
ap
proach
is
t
o
pr
e
ve
nt
viral
tra
nsmi
ssion
a
nd
in
fecti
on.
F
or
thi
s,
me
dical
r
oute
s
are
pr
i
or
it
iz
ed b
y dem
ogra
ph
ic
gro
upin
gs
[17]
.
Delays
in
ide
nt
ifying
an
d
reac
ti
ng
t
o
outb
rea
ks
a
re
co
mm
on
i
n
c
urre
nt
in
f
ect
iou
s
disease
m
on
it
ori
ng
sy
ste
ms
due
to
a
nee
d
for
proacti
ve
respo
nse
capa
bili
ti
es
and
real
-
ti
me
data
integ
rati
on.
A
s
ys
te
m
t
ha
t
us
es
deep
le
arn
i
ng
and
i
nter
net
of
thin
gs
(
IoT
)
te
chnolo
gy
is
nee
ded
t
o
id
entify
i
nf
ect
io
us
disease
out
br
ea
ks
qu
ic
kly
,
i
nter
ve
ne
pr
omptl
y,
and
al
locat
e
re
so
urces
e
ff
ect
i
vely.
This
s
ys
t
em
w
ou
l
d
hel
p
to
reduce
the
i
mp
act
of p
a
ndemic
s.
This
pap
e
r
pr
opos
es
usi
ng
I
oT
se
n
s
ors
f
or
real
-
ti
me
i
nf
e
ct
iou
s
il
lness
monit
or
i
ng
in
con
j
un
ct
io
n
with
dee
p
le
ar
ning
meth
od
s
,
na
mely
rec
urren
t
ne
ur
al
ne
tworks
(
R
NNs
)
.
Tha
nks
t
o
t
his
c
onnecti
on
,
the
sy
ste
m
ca
n
now
exa
mine
more
ty
pes
of
dat
a
and
spot
ir
re
gu
la
riti
es
that
migh
t
be
si
gn
s
of
dis
ease
outbrea
ks
.
The
pro
po
s
ed
meth
od
ma
y
early
predict
infecti
ous
dis
ease
outb
reaks
by
us
in
g
RN
Ns
to
ide
ntify
small
te
mp
oral
tren
ds
in
healt
h
dat
a
gathe
red
fro
m
IoT
de
vices
.
Tech
nolo
gy
can
al
so
pr
e
di
ct
wh
e
n
il
lnes
ses
will
sp
rea
d,
al
lo
wi
ng
prom
pt
pre
ven
ta
ti
v
e
ma
na
geme
nt.
By
c
ombinin
g
geog
raphical
data
f
rom
I
oT
de
vic
es,
the
sy
ste
m
can
pe
rform
s
patia
l
a
nalysis,
pinpo
i
nt
ho
ts
pots,
an
d
f
ollow
in
fect
ed
pe
op
le
wh
e
rev
e
r
t
he
y
perf
orm.
With this
expe
rtise
, w
e
can
bet
te
r
unde
rstan
d
the
dy
namics
of
disease s
pr
e
ad
a
nd r
e
s
pond
w
it
h p
recisi
on.
The
pro
pose
d
syst
em
e
na
bl
es
real
-
ti
me
monit
or
i
ng
of
healt
h
in
dice
s
an
d
il
lness
tre
nd
s
via
con
ti
nu
ous
dat
a
colle
ct
ing
a
nd
a
nalysis
.
Be
cause
of
this,
public
healt
h
offici
al
s
may
qu
ic
kly
res
po
nd
by
dep
l
oy
i
ng
me
di
cal
resour
ces
to
impact
e
d
lo
cat
ion
s,
c
ondu
ct
ing
ta
r
geted
te
sti
ng
,
a
nd
t
ra
ckin
g
c
on
ta
ct
s.
The
pro
po
se
d
s
ys
te
m
e
nhances
public
healt
h
pr
e
par
at
io
n
a
nd
res
pons
e
ca
pacit
ie
s,
wh
ic
h
offe
rs
ti
mely
a
nd
act
ion
able
inf
ormat
io
n
in
the
face
of
in
fecti
ou
s
disease
e
pid
emic
s
.
I
n
the
e
nd,
it
help
s
le
ssen
the
e
f
fect
of
pandemics
a
nd
save
li
ves
by
al
lo
wing
preem
ptive
i
nter
ven
ti
on
m
easur
e
s,
al
loc
at
ing
res
ource
s,
a
nd
disseminati
ng ti
mely pu
blic h
eal
th w
a
rn
i
ng
s
.
Cl
us
te
r
anal
ys
is
in
medici
ne
and
monit
ori
ng
sy
ste
ms
le
d
t
o
early
disease
epidemic
ide
nt
ific
a
ti
on
.
It
pro
vid
es
a
dise
ase
outb
rea
k
s
urveil
la
nce
s
yst
em
us
i
ng
a
n
a
l
ph
a
s
hap
e
a
nd
a
uniq
ue
de
ns
it
y
e
ntr
opy
cl
us
t
erin
g
method
ology
[
18]
.
Infecti
ous
il
lnesses
threa
te
n
p
ub
li
c
heal
th,
ma
king
m
onit
or
i
ng
a
nd
outb
reak
c
oor
din
at
ion
cru
ci
al
.
With
machine
le
ar
ni
ng
(
ML
)
a
nd
data
mini
ng,
our
te
ch
niqu
e
pro
vid
e
s
a
com
plete
an
s
wer
f
or
con
te
m
pora
ry
healt
hcar
e
.
Thi
s
a
ppro
a
ch
im
pro
ves
disease
m
on
it
or
in
g
a
nd
ma
nag
e
ment
usi
ng
sup
port
vect
or
machine
s
(
S
V
M
)
,
ra
ndom
f
or
est
,
a
nd
k
-
mean
s
cl
ust
eri
ng
[19
]
.
I
nf
ec
ti
ou
s
disease
ou
t
br
ea
k
ide
nt
ific
at
ion
helps
public
he
al
th
pro
fessio
nals
react
qu
ic
kly
to
seve
re
public
he
al
th
pr
ob
le
m
s.
Howe
ver,
il
lness
epi
demics
are
s
om
et
imes
inv
isi
ble
[
20]
.
N
oises
from
ordina
r
y
beh
a
vioral
patte
r
ns
an
d
e
xcep
ti
onal
occ
urre
nce
s
may
ham
per
epi
de
mic
m
on
it
ori
ng
s
ys
te
ms
.
Most
detect
ion
appr
oach
es
use
ti
me
series
filt
ering
an
d
s
ta
ti
sti
ca
l
monit
or
i
ng.
T
he
rise
of
C
O
VID
-
19
a
nd
de
ngue
i
n
Ba
ngla
des
h
e
mph
asi
zes
the
nee
d
f
or
a
dig
it
al
healt
h
strat
egy
inte
grat
ing
IoT
a
nd
arti
fici
al
intel
lig
ence
(
AI
)
-
bas
ed
te
c
hnolog
y,
healt
h
be
ha
vio
rs
,
a
nd
at
ti
tud
es
t
o
antic
ipate
,
pr
e
ven
t,
an
d
man
age
the
se
in
fe
ct
iou
s
il
lnesse
s
[
21]
.
Ba
ngla
des
h
m
us
t
e
nhance
pr
e
ve
ntio
n
a
nd
con
t
ro
l.
A
I
or
Io
T
-
based
ne
w
te
chnolo
gies
may
al
s
o
gathe
r
data
an
d
fore
cast
fu
t
ur
e
eve
nts,
maki
ng
it
simpl
e
for
the
he
al
th
minist
ry
t
o
ma
ke
ef
fici
ent
de
ci
sion
s.
Most
infecti
ous
il
lnes
ses
sp
r
ead
qu
i
ckly
a
nd
ha
ve
a
wide
eff
ect
.
O
nce
t
hey
sprea
d,
t
he
y
will
infect
a
broa
d
reg
i
on,
posin
g
se
rio
us
healt
h
a
nd
s
ecur
it
y
haza
rd
s
[
22]
.
Ther
e
f
or
e,
earl
y
in
fecti
ous
il
lness
s
urveil
la
nc
e
an
d
pr
e
ven
t
ion
a
re
c
ru
ci
al
.
Cu
rr
e
nt
m
on
i
toring
m
et
hods
can
antic
ipate
the
occ
urre
nce
of
in
fecti
ous
il
lness
es.
Se
nso
r
data
needs
to
be
more
di
ver
se,
acc
ur
at
e,
a
nd
com
plete
,
mak
ing
m
onit
or
i
ng
fin
dings
dif
ficult
.
Mo
nitori
ng
s
ys
te
ms
c
annot
quic
kl
y
handle
the
gro
wing
amo
un
t
of
data due to
li
mit
ed l
ocal res
ources
.
To
a
ddres
s
t
his
pu
blic
healt
h
c
oncer
n,
cut
ti
ng
-
e
dge
te
c
hnol
ogy
li
ke
de
ep
le
a
rn
i
ng
is
nee
de
d.
It
examine
s
w
hat
disease
outbre
aks
are
rele
vant
now
,
how
we
ll
dee
p
le
ar
ni
ng
meth
ods
ha
ve
bee
n
done
in
earl
y
detect
ion,
a
nd
ho
w
t
o
unde
rstan
d
releva
nt
dee
p
le
a
rn
i
ng
te
c
hn
i
qu
e
s
[
23]
pro
pe
rly.
Non
-
phar
mace
utica
l
appr
oach
es
to
con
t
ro
l
resp
i
ra
tor
y
disease
s
l
ike
the
flu
a
re
a
sig
nificant
le
arn
in
g
from
the
epi
demic.
Also
,
few
e
r
hos
pital
admissio
ns
or
repor
ts
may
not
mea
n
fe
wer
resp
i
ratory
di
sease
age
nts.
Howe
ver,
go
ve
rnment
awar
e
ness
init
ia
ti
ves
a
nd
publ
ic
co
op
e
rati
on
ha
ve
dr
ast
ic
al
ly
reduce
d
dis
ease
s
pr
ea
d
[
24]
.
More
over
,
public
healt
h
init
ia
ti
ves
aff
ect
di
ff
e
ren
t
disease
s
diff
e
re
ntly.
I
de
ntify
i
ng
t
he
inter
ven
ti
ons
t
hat
ha
ve
the
gr
eat
est
influ
e
nce
on
disease tra
ns
miss
ion
re
qu
ire
s fu
rther resea
rc
h.
The
I
oT
an
d
massi
ve
da
ta
analysis
in
he
al
thcare
ha
ve
captu
red
data
forme
rly
do
ne
ma
nual
ly.
Kno
wled
ge
a
nd
real
-
ti
me
m
on
it
ori
ng
a
re
need
e
d
t
o
dia
gnose
an
d
st
op
infecti
ous
diseases.
E
ven
i
n
distant
reg
i
on
s
,
the
I
oT
has
capt
ur
e
d
real
-
ti
me
se
ns
or
y
data
f
rom
pe
ople
,
heal
th
s
ys
te
ms,
an
d
ec
osystems
[25]
.
It
migh
t
offe
r
preven
ti
ve
act
io
ns
util
iz
ing
IoT
net
work
data
an
d
e
valuate
their
im
pleme
ntati
on
.
I
n
pa
ndemic
s
li
ke
CO
VID
-
19,
monit
ori
ng
the
disease'
s
tr
ansmissi
on
an
d
et
iolo
gy
is
c
ru
ci
al
,
par
ti
c
ularly
i
n
poor
na
t
ions
[26]
.
Re
al
-
ti
m
e
data
is
tr
ack
ed
usi
ng
a
m
obil
e
app
an
d
web
sit
e.
O
ur
s
ys
te
m
is
str
onger
a
nd
m
or
e
eff
ic
ie
nt
than othe
rs.
T
he
n
ovel
coro
na
vir
us
S
ARS
-
C
ov
-
2
has
tri
gg
e
red
C
OVID
-
19, a
global epi
de
mic
[27]
. Coughin
g,
sh
ort
ne
ss
of
br
eat
h,
an
d
hi
gh
te
mp
erat
ur
e
a
r
e
f
re
qu
e
nt
sym
pto
m
s.
CO
VID
-
19
i
ns
ta
nces
are
gro
wing,
making
man
ual
i
den
ti
f
ic
at
ion
of
c
onta
gious
pe
ople
in
public
set
ti
ng
s
dif
ficult
[
28]
.
T
his
sy
st
em
is
e
valuate
d
on
a
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
1
175
-
1
186
1178
real
-
ti
me
datas
et
an
d
a
n
existi
ng
s
neeze
-
c
ou
gh
dataset
.
Te
s
ts
re
veal
that
the
s
ugge
ste
d
te
chn
i
qu
e
ou
t
pe
rforms
innov
at
ive
met
hods
[
29]
.
A
deep
fee
dfo
r
ward
mu
lt
il
aye
r
pe
rce
ptr
on
A
I
c
hatb
ot
inter
act
ion
a
nd
pre
dicti
on
model
and
vac
uum
in
theo
reti
cal
and
pr
act
ic
al
in
structio
ns
f
or
desig
ning
A
I
c
hatb
ots
f
or
li
fe
style
impro
ve
ment
pro
grams
[30]
.
Our
sug
gested
m
odel
's
ti
me
com
plexity
a
nd
te
sti
ng
acc
uracy.
It
d
isc
us
s
es
me
dical
c
ha
tbo
ts'
f
unct
ions,
us
es
,
and
iss
ues
the
y
offe
r
duri
ng
healt
h
e
mer
ge
ncies
li
ke
pa
nd
e
mics.
Ge
ne
ral
inte
rn
ist
s
treat
m
os
t
i
nfect
iou
s
il
lnesses.
H
owever,
i
nf
ect
io
us
disease
i
nter
ni
sts
help
i
den
ti
fy
a
nd
treat
se
ver
e
,
unusual
,
or
c
omplex
in
f
ec
ti
on
s
[31]
.
An
ti
micr
ob
ia
l
me
dicin
es,
a
ntibioti
c
tolera
nce,
im
m
un
iz
at
io
n,
ot
he
r
ca
us
es,
an
d
the
pr
e
valenc
e
an
d
cl
inica
l
pr
esen
ta
ti
on
of
infec
ti
ou
s,
viral,
f
ungal,
a
nd
pa
ra
sit
e
diseases
are
essenti
al
to
infecti
ous
dis
ease
mana
geme
nt
[
32]
.
T
he
inf
ect
iou
s
disease
pr
act
ic
e
has
num
erous
m
odel
s.
Infecti
ous
dise
ase
phys
ic
ia
ns
ma
y
pr
act
ic
e
inf
ect
iou
s
il
lnesses
in
a
cl
inic
or
al
ongs
ide
normal
me
dic
al
care.
Like
man
y
il
lnesse
s,
early
identific
at
ion
may
help
patie
nts
get
t
he
c
orrect
treat
me
nt
to
re
duce
harm
or
is
olate
th
em
to
prev
e
nt
sp
rea
d
[33]
.
I
n
this
ca
se,
co
mputer
i
ntell
igence
ca
n
forecast
patie
nt
infecti
on
ris
k
an
d
al
ert
me
dical
perso
nnel
to
act
qu
ic
kly
.
It
e
xp
l
or
es
i
nf
ect
iou
s
il
lness
M
L
a
pp
li
cat
io
ns
.
Our
main
disea
se
c
on
cern
s
a
re
dia
gnos
is,
transmissi
on,
treat
ment
re
spo
ns
e,
a
nd
res
il
ie
nce
[
34]
.
High
values
ma
y
be
of
releva
nce
f
or
in
fecti
ous
disease
researc
h.
M
ach
ine
-
le
ar
ning
m
od
el
s
f
or
earl
y
or
real
-
ti
me
ep
idemic
ver
ific
a
ti
on
a
re
a
ne
w
and
in
novative
fiel
d
with ma
ny
st
udy met
hodolo
gi
es, mak
i
ng
it
diff
ic
ult t
o
co
mp
a
re s
tu
dies
method
ologica
ll
y,
ev
e
n
th
ough almo
st
al
l
agr
ee
that
major
in
fecti
ous
disease
outbrea
ks
can
be
monit
ored
[
35]
.
T
his
im
plies
that
M
L
m
igh
t
be
extensi
vely
a
ppli
ed
in
pu
blic
healt
h
a
nd pre
ven
ta
ti
ve
ef
forts.
2.
PROP
OSE
D MET
HO
D
Weara
ble
se
nsors
,
s
mart
th
er
mo
mete
rs,
an
d
smart
phones
w
it
h
healt
h
monit
or
i
ng
ap
ps
a
re
s
om
e
I
oT
dev
ic
es
us
e
d
to
gathe
r
real
-
ti
me
healt
h
data.
I
n
real
-
ti
me,
these
dev
ic
es
r
ecord
c
ru
ci
al
healt
h
data,
s
uc
h
as
vital
sign
s,
sympto
ms,
a
nd
more.
Data
str
eams
are
sa
fel
y
se
nt
to
cent
ral
ser
ver
s
or
cl
oud
platf
orm
s
for
ad
diti
onal
proc
essing
or
ag
gregati
on.
T
he
fi
nal
dataset
wil
l
li
kely
inclu
de
a
broa
d
ra
ng
e
of
data
f
rom
s
ever
al
so
urces
.
Data
streams
are
preprocesse
d
as
the
y
reac
h
c
entrali
zed
se
rvers
or
cl
oud
platfo
rms
to
pro
vid
e
consi
ste
ncy
a
nd
qual
it
y
of
the
data.
First,
it
cl
eans
the
data
by
rem
ov
ing
no
ise
a
nd
missi
ng
num
be
rs.
It
normali
zes
it
so
t
hat
it
is
c
on
sist
e
nt
acr
oss
al
l
our
s
our
ces.
T
he
oper
at
ion
s
of
pr
e
proces
sin
g
a
nd
stora
ge
pro
vid
e t
he gr
ound
work f
or th
e f
ollow
i
ng ana
lyses a
nd d
eci
sion
s
.
An
al
yzin
g
se
quentia
l da
ta
ge
ner
at
e
d
f
rom IoT
dev
ic
es
us
e
s d
ee
p
le
ar
ning
models,
namel
y
RN
Ns
a
nd
var
ia
ti
ons
s
uc
h
as
LS
TM
dataset
s.
Tem
pora
l
patte
rns
s
ugge
sti
ve
of
i
nf
ect
iou
s
disease
outb
reak
s
are
le
arn
e
d
by
trai
ni
ng
t
he
se
models
usi
ng
histo
rical
data.
Disease
dynamics
ma
y
be
bett
er
under
st
ood
us
in
g
dee
p
le
arn
in
g
mod
el
s,
ide
ntify
i
ng
im
portant
aspects
li
ke
t
emp
or
al
patte
rn
s
an
d
outl
ie
rs
f
r
om
t
he
data.
In
c
orp
or
at
in
g
geo
l
ocati
on
da
ta
colle
ct
ed
f
rom
Io
T
dev
ic
es
into
t
he
anal
ytic
f
rame
wor
k
m
ay
bette
r
un
derst
and
the
ge
ogra
phic
al
patte
rns
of
disease
tra
ns
m
issi
on
.
S
patia
l
analyti
c
meth
od
s
are
use
d
to
pinpo
i
nt
dis
eas
e
epicenters
,
stu
dy
tra
nsmi
ssio
n
patte
r
ns
,
a
nd
f
ollow
i
nf
ec
te
d
pe
op
le
's
w
her
ea
bouts.
T
o
hel
p
public
healt
h
offici
al
s
ma
ke
e
du
cat
e
d
dec
isi
on
s,
visu
al
i
zat
ion
te
ch
no
l
og
ie
s
li
ke
hea
t
maps
an
d
GIS
s
of
t
war
e
s
how
th
e
geog
raphical
di
stribu
ti
on
of
i
nf
ect
io
us
il
lnes
ses.
T
he
te
ch
nolo
gy
us
es
ML
al
gorith
ms
t
o
forecast
t
he
s
pr
ead
of
infecti
ous
dise
ases
a
nd
ide
nt
ify
outbrea
ks
earl
y.
W
he
n
data
stream
s
sho
w
unusua
l
be
hav
i
or,
a
nomaly
detect
ion
al
go
r
it
hm
s
will
ra
ise
the
re
d
fl
ag
a
nd
noti
f
y
public
healt
h
offici
al
s
of
a
possible
e
pi
demic.
Pr
e
dicti
ve
m
odel
s
us
e
hist
ori
cal
data
an
d
real
-
ti
me
in
puts
t
o
a
ntici
pa
te
the
f
uture
co
urse
of
di
sease
transmissi
on
a
nd
e
na
ble
pro
act
ive
interve
nt
ion
strat
egies
.
Be
fo
re
a
n
in
f
ect
iou
s
diseas
e
epidemic
wor
sens,
sever
al
ste
ps
ar
e
ta
ken
to
le
sse
n
it
s
ef
fect,
su
c
h
as
al
locat
ing
resou
rces,
de
ve
lop
in
g
co
ntainme
nt
met
hods,
a
nd
commu
nicat
in
g
with
sp
eci
fic
po
pu
la
ti
ons
via
publi
c
healt
h
ca
mp
ai
gns
.
T
he
pro
pose
d
s
ys
te
m's
interco
nnect
io
n parts a
re sh
own
in Fi
gure
2
blo
c
k diag
ram.
Figure
2. Bl
oc
k diag
ram of
prop
os
ed
integ
r
at
ed
in
fecti
ous
disease s
urveil
la
nce s
ys
te
m
wi
th IoT a
nd
deep le
ar
ning
D
a
t
a Co
l
l
ec
t
i
o
n
(Io
T
D
ev
i
ce
s
)
D
a
t
a Pre
p
r
o
ce
s
s
i
n
g
&
In
t
eg
r
at
i
o
n
D
ee
p
L
ear
n
i
n
g
A
n
al
y
s
i
s
(R
N
N
s
)
G
e
o
s
p
a
t
i
al
A
n
al
y
s
i
s
E
ar
l
y
D
e
t
ec
t
i
o
n
&
Pred
i
c
t
i
o
n
Real
-
T
i
me M
o
n
i
t
o
ri
n
g
& Res
p
o
n
s
e
Pu
b
l
i
c
H
eal
t
h
D
eci
s
i
o
n
Su
p
p
o
r
t
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
omp E
ng
IS
S
N:
20
88
-
8708
Deep l
ear
ning
for infect
io
us
di
sease
s
ur
vei
ll
an
ce
integ
ra
ti
ng
inter
net
of th
ing
s
…
(
Subra
manian
Sumith
ra
)
1179
2.1.
Io
T
d
e
vices f
or
s
u
rveil
lance
The
pr
opos
e
d
sy
ste
m
m
os
tl
y
us
es
f
our
mai
n
ki
nds
of
I
oT
de
vices
t
o
colle
ct
data
an
d
m
onit
or
healt
h
factors.
T
hese
dev
ic
es
ca
n
id
entify
i
nf
ect
io
us
diseas
e
outb
reak
s
ea
rly
a
nd
prov
i
de
real
-
ti
me
surveil
la
nc
e
data.
Four
of the
m
ost
co
m
mon
IoT
d
e
vices are:
a.
Weara
ble
sens
or
s
:
t
he
m
os
t
popula
r
wear
a
bl
e
se
ns
ors
on
th
e
I
oT.
When
w
orn
on
th
e
bo
dy,
these
m
onit
or
s
const
antly
m
onit
or
metri
cs
li
ke
hear
t
rate,
t
empe
rature,
a
nd
res
pirati
on
r
at
e.
I
ns
ig
hts
on
pe
op
le
's
heal
th
and
ide
ntif
ying
a
bnormal
it
ie
s
that
may
in
dicat
e
in
fecti
ons
may
be
gai
ned
f
r
om
thes
e
te
sts.
Weara
ble
sens
or
s
pro
vide
con
ti
nu
ou
s
he
al
th
data
with
ou
t
bein
g
obtr
us
ive
,
w
hich
i
s
a
major
ben
e
fit.
The
y
ma
y
be
seaml
essly i
ntegr
at
e
d
int
o
e
ve
ryda
y
act
ivit
ie
s.
b.
Smart
the
rm
ome
te
rs:
t
he
te
c
hnolog
y
al
so
r
el
ie
s
on
sma
rt
therm
om
et
e
rs,
wh
ic
h
can
acc
ur
at
el
y
mea
sur
e
a
patie
nt's
te
m
pe
ratur
e
.
With
th
e
help
of
the
I
oT
,
t
hermo
mete
rs
li
ke
t
his
m
ay
wirelessl
y
s
end
te
mp
e
ratu
r
e
read
i
ng
s
to
m
ai
n
se
rv
e
rs
or
cl
oud
platf
or
m
s,
al
lo
wing
for
anal
ys
is
a
nd
monit
or
i
ng
in
real
-
ti
me.
Sma
rt
therm
om
et
e
rs,
wh
ic
h
record
pa
tt
ern
s
of
fluct
uating
co
re
te
mp
e
ratur
e
s,
migh
t
help
doct
or
s
dia
gnos
e
fe
bri
le
diseases a
nd
possible i
nf
ect
io
us
disease e
pide
mics fa
r
ea
rlie
r.
c.
Smart
phones
with
he
al
th
m
on
it
ori
ng
a
pps
:
nowad
a
ys,
mo
st
people
ha
ve
acce
ss
t
o
smartp
hone
s
with
instal
le
d
sen
s
or
s
,
s
uch
as
GP
S
,
gyrosco
pes,
a
nd
acce
le
ro
mete
rs.
T
he
se
sens
ors
a
r
e
us
e
d
by
he
al
th
monit
or
i
ng
ap
plica
ti
on
s
t
o
re
cord
the
am
ount
of
phys
ic
al
act
ivit
y,
sle
ep
durati
on,
a
nd
moveme
nt
patt
ern
s
of
us
e
rs.
Disea
se
m
on
it
ori
ng
and
outb
rea
k
r
esp
on
se
init
ia
ti
ves
ca
n
be
nef
it
great
ly
from
data
pr
ov
i
ded
by
app
li
cat
io
ns
t
hat
al
low
us
e
rs
to
repo
rt
s
ymptom
s
or
t
race
co
ntact
s.
With
s
o
ma
ny
pe
ople
us
i
ng
smartp
hone
s,
the
sur
veill
ance
sy
ste
m
can
re
ach
m
or
e
peop
le
and
get
the
m
involve
d,
m
akin
g
it
bette
r
at
sp
otti
ng a
nd tr
ackin
g
in
fecti
ous il
lnesses
.
d.
En
vironm
e
nta
l
sens
ors
:
I
oT
de
vices
c
onnect
ed
with
e
nviro
nm
e
ntal
sen
sor
s
great
ly
as
sist
in
m
on
it
or
in
g
ai
r
qu
al
it
y,
humid
it
y
le
vels,
a
nd
ot
her
e
nv
i
ron
mental
par
a
m
et
ers
relat
ed
to
disease
tra
nsmi
ssio
n.
T
he
se
sens
or
s
may
de
te
ct
ai
rb
orne
i
nf
ect
io
ns
,
c
onta
minants
,
an
d
env
i
r
onme
ntal
factors
that
pr
omote
the
spre
ad
of
resp
i
rato
ry
vir
us
es.
P
ubli
c
healt
h
offici
al
s
may
bette
r
unde
rstan
d
e
nv
i
ronme
ntal
risk
factors
an
d
how
env
i
ronme
ntal
conditi
ons
af
fe
ct
disease
tra
nsmi
ssion
dyna
m
ic
s
by
i
ncor
por
at
ing
data
fro
m
en
vir
onment
al
sens
or
s
into
th
e
m
on
it
ori
ng
s
ys
te
m.
Fig
ure
3
flo
wch
a
rt
s
hows
the
RN
N
par
t
of
the
s
ys
te
m
in
act
ion,
f
ro
m
data pre
par
at
io
n
to
r
eal
-
ti
me
re
act
ion
a
nd ale
rt act
ivati
on d
e
pendin
g o
n
R
N
N ou
t
pu
t.
Figure
3. Flo
w
char
t
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
1
175
-
1
186
1180
e.
RNN
i
n
disea
se
surveil
la
nce
:
f
or
te
m
poral
data
patte
rn
analysis
a
nd
t
he
ease
of
ta
s
ks
li
ke
a
noma
ly
detect
ion an
d p
red
ic
ti
ve
a
naly
sis, the
pro
pos
ed
s
ys
te
m
de
pe
nd
s
sig
nifica
ntly
on RN
Ns. Th
e u
se
of R
NN
s:
1)
Temp
or
al
data
analysis
: R
NNs excel at
hand
li
ng
se
quentia
l
data,
wh
ic
h
ma
kes
the
m
perfe
c
t for e
xamini
ng
tren
ds
in
th
e
da
ta
acqu
ire
d
from
Io
T
de
vice
s
over
ti
me.
A
cru
ci
al
s
kill
fo
r
unde
rstan
di
ng
t
he
co
urse
of
infecti
ous il
lne
sses is the
capa
ci
ty for
R
N
Ns t
o
rec
ord de
pe
nd
e
ncies a
nd tr
ends
ov
e
r
ti
me.
2)
Pr
e
dicti
ve
a
nalysis
:
R
NN
s
a
r
e
use
d
to
foretel
l
fu
t
ur
e
c
hanges
an
d
tre
nd
s
by
anal
yzin
g
past
data.
RN
N
s
can
pr
e
dict
f
ut
ur
e
ou
t
br
ea
ks
or
pi
npoin
t
l
ocati
ons
pron
e
to
in
fecti
on
by
anal
yzin
g
t
rends
i
n
disease
transmissi
on
a
nd
patie
nt
heal
th
metri
cs
.
Due
to
t
he
ca
pacit
y
f
or
pre
dicti
on,
pr
e
ve
ntati
ve
act
ion
s
ma
y
be
ta
ken
t
o
le
sse
n t
he
se
ver
it
y o
f i
nf
ect
io
us
dise
ase o
utbreak
s.
3)
Anomal
y
detec
ti
on
:
a
no
t
her u
sage
f
or
RN
Ns
is
a
nomaly
d
et
ect
ion
,
w
hich
i
nvolv
e
s f
in
ding d
at
a points
that
don't
fit
the
nor
m.
R
NNs
m
ay
detect
the
e
mer
gen
ce
of
i
nf
ect
io
us
disea
ses
by
c
onsta
n
tl
y
scan
ning
da
ta
streams
f
rom
Io
T
se
nsors
and
al
erti
ng
us
ers
t
o
an
y
de
viati
ons
from
normal
healt
h
metri
cs
or
env
i
ronme
ntal
ci
rcu
msta
nce
s.
It
is
possi
ble
to
inter
ve
ne
an
d
li
mit
the
sit
uation
more
quic
kl
y
i
f
abno
rmali
ti
es are
detect
ed
ea
rly.
4)
Re
al
-
ti
me
mon
it
or
in
g
a
nd
res
pons
e:
disease
dy
namics
a
nd
patie
nts'
healt
h
sta
te
s
ma
y
be
trac
ked
in
re
al
ti
me
us
in
g
RN
Ns.
RN
Ns
a
nal
yze d
at
a
in
real
-
ti
me,
al
lo
wing
f
or
quic
k
react
ion
s
to n
e
w
ris
ks
by p
r
oduci
ng
warnin
gs
or
tri
gg
e
rin
g
inte
rv
e
ntion
proce
dur
es
in
res
pons
e
to
ab
normali
ti
es
or
t
he
outc
ome
s
of
pre
dicti
on
analyses
.
5)
Ad
a
ptive
le
ar
ni
ng
:
RN
Ns
ca
n
evo
l
ve
an
d
m
od
i
fy
t
heir
inte
rn
al
re
presenta
ti
on
s
in
res
ponse
to
f
resh
i
nput,
wh
ic
h
e
na
bles
them
to
gr
a
du
al
ly
e
nh
a
nc
e
their
ca
pacit
y
to
ide
ntif
y
an
om
al
ie
s
an
d
make
acc
urat
e
pr
e
d
ic
ti
on
s
.
T
he
sy
ste
m's
co
nt
inu
e
d
ef
ficacy
in
eve
r
-
c
ha
ng
i
ng
infecti
ou
s
il
lness
set
ti
ngs
i
s
guara
nteed
by
this
ada
ptive
l
earn
i
ng
featu
re
.
Fig
ur
e
3
e
xpla
ins
the
Fl
ow
c
har
t
of
pro
pos
ed
mec
ha
nism.
In
it
ia
ll
y,
c
ollec
ts
the
data
f
r
om
I
oT
a
nd
se
nsor
dev
ic
es
.
Ne
xt
do
i
ng
t
he
data
pr
e
-
pr
ocessin
g
and
sepa
rate
the
Tem
poral
da
ta
seq
uen
ci
ng
an
d
f
eat
ur
e
ext
ra
ct
ion
performa
nce.
The
RN
N
model
is
us
e
d
to
trai
ni
ng
an
d
validat
e
t
he
da
ta
.
Fu
rt
hermo
re,
it
integ
rates
with
t
he
real
ti
me
monit
or
i
ng
a
nd
finall
y
it
f
or
wards
the
real
-
ti
me
res
pons
e
and
al
ert act
ivati
on.
3.
RESU
LT
S
AND DI
SCUS
S
ION
The
ou
tc
om
es
and
a
nalys
es
of
the
pro
pos
ed
in
fecti
ou
s
disease
sur
veill
ance
sy
ste
m
highli
gh
t
it
s
eff
ic
acy
i
n
usi
ng
IoT
an
d
RN
N
te
ch
no
l
og
ie
s
for
proacti
ve
r
esp
on
se
t
o
dise
ase
ou
t
br
ea
ks,
early
ide
ntific
a
ti
on,
and
real
-
ti
me
monit
or
i
ng.
T
he
s
ys
te
m
can
gat
her
va
rio
us
healt
h
in
dic
at
or
s
an
d
e
nviro
nm
e
ntal
el
e
ments
per
ti
ne
nt
t
o
di
sease
tra
ns
mi
ssion
by
c
ollec
ti
ng
data
fro
m
Io
T
dev
ic
e
s,
s
uc
h
as
we
arab
le
sen
sors
an
d
env
i
ronme
ntal
monit
or
s
.
In
c
orp
or
at
in
g
R
N
N
s
al
lows
for
e
xa
minin
g
data
tr
ends
ov
e
r
ti
me
,
al
lo
wing
for
more
accurate
pre
dicti
ve
m
odel
ing
an
d
ide
ntify
i
ng
outl
ie
rs.
By
pr
e
dicti
ng
dise
ase
ou
t
br
ea
ks
and
s
pott
ing
unusual
patte
rn
s
t
hat
m
igh
t
in
dicat
e
ne
w
ris
ks
,
t
he
a
ppr
oach
s
hows
promise
wh
e
n
pu
t
int
o
p
racti
ce.
The
te
c
hnol
ogy
eff
ic
ie
ntly
noti
fies
healt
hca
re
auth
or
it
ie
s
of
a
ny
outb
reaks
by
m
onit
or
i
ng
r
eal
-
ti
me
data
s
treams,
al
lo
wi
ng
f
or
ti
mely inter
vent
ion
a
nd contai
nm
e
nt meas
ure
s.
RNNs'
ada
ptiv
e
le
arn
i
ng
cap
abili
ti
es
make
the
s
ys
te
m
s
m
arter
a
nd
faste
r
as
it
le
arns
f
r
om
ne
w
data
and
updates
it
s
pre
dicti
on
m
od
el
s
.
T
he
sys
te
m's
re
voluti
onar
y
pote
ntial
to
prov
i
de
a
proacti
ve
a
pproach
t
o
public
healt
h
mana
geme
nt
a
nd
m
oder
nize
i
nf
ect
io
us
disea
se
m
on
it
ori
ng
has
bee
n
e
mph
asi
zed
in
discu
ssion
s
.
The
sy
ste
m
protect
s
the
publ
ic
's
healt
h
a
nd
safet
y
by
al
lo
wing
healt
hca
r
e
offici
al
s
t
o
f
or
esee
a
nd
le
s
sen
t
he
eff
ect
s
of
co
nt
agious
il
lness
es
via
t
he
use
of
RN
N
a
nd
the
I
oT.
O
ngoing
im
pro
veme
nt
an
d
e
thica
l
consi
der
at
io
ns
thr
ough
out
s
yst
em
dev
el
opment
a
nd
de
plo
yme
nt
are
ne
ces
sar
y
due
t
o
obsta
cl
es,
inc
lud
in
g
data
pr
iva
cy
issues
an
d
al
go
rithmic
biases.
It
highli
ghts
how
t
he
pro
po
sed
s
ys
te
m
ca
n
revoluti
oniz
e
ho
w
infecti
ous
dise
ase res
pons
e
a
nd m
on
it
ori
ng
are c
onduct
ed
i
n
the
futu
re.
3.1.
R
N
Ns
d
atase
t
Table
1
prov
i
des
a
co
mpre
hensi
ve
datase
t
fo
r
run
ning
an
infecti
ou
s
disease
m
on
it
or
i
ng
s
ys
te
m
inco
rpor
at
in
g
RNNs
a
nd
I
oT
te
chnolo
gies
.
A
A
ti
mesta
m
ps
c
orrespo
nd
i
ng
t
o
eac
h
r
ow
in
t
he
ta
ble
br
ie
fly
sh
ow
the
ma
ny
en
vir
onme
ntal
fact
or
s
a
nd
he
al
th
in
dicat
ors
li
n
ked
t
o
patie
nts.
A
ddin
g
a
“
Pati
ent
I
D
”
c
olu
m
n
al
lows
f
or
t
he
longit
ud
i
nal
tr
ackin
g
of
va
ri
ou
s
patie
nts'
he
al
th
record
s,
al
lowing
f
or
more
ta
rg
et
e
d
analysis
and
m
onit
or
i
ng.
E
nv
ir
onme
nt
al
var
ia
bles,
s
uch
as
te
m
per
a
ture
an
d
ai
r
qu
al
it
y
inde
x,
pro
vid
e
bac
kgr
ou
nd
that
may
im
pact
he
al
th
ou
tc
om
es
.
M
ean
w
hile,
phys
i
ologica
l
m
easur
e
s,
s
uch
a
s
hear
t
rate
,
bo
dy
te
m
pe
rature
,
an
d
br
eat
hing
rate,
pr
ov
i
de
insig
hts
i
nto
patie
nts'
phys
iol
og
ic
a
l
co
ndit
ion
.
T
o
de
te
ct
outl
ie
rs
a
nd
pr
e
dict
di
sease
ou
t
br
ea
ks
,
it
is
cr
ucial
to
a
na
lyze
tre
nds
a
nd
patte
r
ns
over
ti
me,
made
possible
by
the
te
mp
oral
c
omp
on
e
nt
recorde
d
by
t
he
“
Time
sta
mp
”
column.
The
ta
ble
-
enca
psul
at
ed
dataset
is
a
gr
eat
sta
rtin
g
po
i
nt
f
or
in
fe
ct
iou
s
disease s
urveil
la
nce, real
-
ti
me
monit
ori
ng a
nd
respo
ns
e,
pre
dicti
ve
a
nal
ys
i
s,
a
nd RN
Ns
m
od
el
t
rainin
g.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
omp E
ng
IS
S
N:
20
88
-
8708
Deep l
ear
ning
for infect
io
us
di
sease
s
ur
vei
ll
an
ce
integ
ra
ti
ng
inter
net
of th
ing
s
…
(
Subra
manian
Sumith
ra
)
1181
Table
1.
Pati
en
t
healt
h datase
t
Tim
esta
m
p
Patien
t
ID
Hear
t R
ate
(bp
m
)
Bo
d
y
T
em
p
erature
(°
C)
Res
p
iratory
Rate
(breaths
/m
in
)
Env
iron
m
en
tal
Tem
p
e
rature
(°
C)
Air
Qu
ality
Ind
ex
2024
-
05
-
01
0
8
:4
5
:
0
0
001
82
3
6
.8
21
2
2
.8
38
2024
-
05
-
01
0
9
:3
0
:
0
0
002
76
3
6
.7
18
2
2
.6
36
2024
-
05
-
01
0
9
:4
5
:
0
0
003
78
3
6
.8
19
2
2
.7
37
2024
-
05
-
01
1
0
:0
0
:
0
0
004
80
3
6
.9
20
2
2
.8
38
2024
-
05
-
01
1
0
:1
5
:
0
0
005
77
3
6
.6
17
2
2
.5
35
2024
-
05
-
01
1
0
:3
0
:
0
0
006
79
3
6
.7
18
2
2
.6
36
3.
2
.
R
N
Ns
data
se
t
In
t
he
c
on
te
xt
of
i
nf
ect
io
us
di
sease
surveil
la
nce,
Fig
ur
e
4
shows
a
gr
a
ph
c
omparin
g
t
he
numb
e
r
of
repor
te
d
i
ns
ta
nc
es
of
il
lness
with
t
he
num
be
r
of
cases
pr
e
dicte
d
by
t
he
RNNs
m
odel
.
The
y
-
axis
s
hows
t
he
total
nu
m
be
r
of
instances
of
i
ll
ness,
w
hile
the
x
-
axis
s
how
s
the
ti
mefr
am
e,
usual
ly
in
da
ys
or
inter
val
s.
The
gr
a
ph
sho
ws
ob
s
er
ved
il
lne
ss
cases
an
d
forecast
ed
dis
ease
cases
on
two
li
nes
,
w
it
h
each
data
po
i
nt
corres
pondin
g
to
a
gi
ven
ti
m
est
amp.
T
he
two
li
nes
dem
onstrat
e
how
w
el
l
the
R
NN
model
pr
e
dicts
f
uture
il
lness
patte
r
ns.
Wh
e
n
the
pr
e
dicte
d
an
d
real
val
ues
are
nea
r,
the
RN
N
model
ca
ptures
th
e
in
fecti
ous
di
sease
dynamics
and
patte
rn
s
w
el
l.
Figure
4. Com
par
is
on of
a
ct
ua
l
v
s
p
re
dicte
d
d
ise
ase
c
ases
In
sta
nces
wh
e
re
the
m
odel
ov
e
resti
mate
s
or
unde
resti
mate
s
il
lness
pr
e
valence
,
as
sh
ow
n
b
y
discre
pan
ci
es
betwee
n
th
e
li
nes,
in
dicat
e
r
egio
ns
nee
ding
ref
i
neme
nt
a
nd
im
pro
veme
nt
.
The
pe
rfo
rm
ance
of
the
RN
N
mode
l
in
earl
y
detect
ion
,
t
rend f
ore
cast
ing
,
an
d
de
ci
sion
-
maki
ng f
or p
ubli
c
healt
h
init
ia
ti
ves
m
ay
be
evaluate
d
by
sta
keholde
rs
via
visua
l
co
mp
a
r
ison
s
of
act
ual
an
d
e
xp
ect
e
d
il
lness
cases
over
ti
me.
Au
t
horiti
es
may
get
ti
mely
insig
hts
to
re
duce
the
s
pr
ea
d
and
ef
fect
of
i
nfect
iou
s
il
lness
es
by
us
in
g
R
NN
-
ba
sed
pre
di
ct
ive
modeli
ng,
as
s
how
n
by
a
gr
a
ph
s
ho
wing
a
high
degree
of
co
ns
ist
enc
y
be
tween
act
ual
and
antic
ipate
d
value
s
in in
fecti
ous
disease m
on
it
ori
ng.
0
20
40
60
80
100
120
140
160
180
Act
ual D
is
eas
e
C
a
s
es
Predicted
Dis
ease
Cases
Dise
ases
Ca
ses
Ti
me Inst
an
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
1
175
-
1
186
1182
3.
3
.
R
N
Ns
m
odel
p
erf
orm
ance
e
valua
tio
n
Table
2
c
onf
usi
on
matri
x
offe
rs
a
c
omp
reh
e
ns
ive
eval
uatio
n
of
the
R
NN
model'
s
a
bili
ty
to
forecast
infecti
ou
s
il
lness
ou
tc
om
es
.
Wh
e
n
the
m
od
el
co
rr
ect
ly
predict
s
the
e
xistence
of
the
il
lness,
it
is
cal
le
d
a
tr
ue
po
sit
ive
(
TP).
On
the
oth
e
r
ha
nd,
w
hen
th
e
disease
is
pr
es
ent
bu
t
no
t
ide
ntifie
d
by
the
model,
it
is
ca
ll
ed
a
false
ne
gative
(F
N
).
true
ne
ga
ti
ves
(T
N)
show
that
the
m
odel
pro
pe
rly
predict
s
the
a
bs
e
nce
of
t
he
il
lne
ss.
I
n
con
t
rast,
false
po
sit
ives
(
FP
)
sh
ow
that
t
he
model
mist
a
ke
nly
predict
s
th
e
existe
nce
of
the
disease
w
he
n
it
is
abse
nt.
By ass
e
ssing
t
hese
values,
sta
kehold
e
rs
ma
y
ev
al
uat
e the m
od
el
's s
ensiti
vity,
s
pec
ific
it
y,
and acc
ur
ac
y
in
i
den
ti
f
ying
il
lness
cases.
T
his
inf
ormat
io
n
can
the
n
in
fl
uen
ce
decisi
on
-
ma
king
on
pu
blic
healt
h
trea
tments
and meth
ods
f
or su
cce
ssf
ully
all
ocati
ng r
es
ources
to batt
le
inf
ect
io
us
disea
ses.
Table
2
. C
onf
usi
on
m
at
rix
of
RNNs
d
ise
ase
p
re
dicti
on
Actu
al/ pred
icted
c
lass
Predicted
po
sitiv
e (
d
iseas
e)
Predicted
neg
ativ
e
(no
dis
ease)
Actu
al po
sitiv
e (
d
iseas
e)
120
30
Actu
al neg
ativ
e (
n
o
dis
ease)
20
850
As
s
how
n
i
n
F
igure
5,
t
he
R
NN
m
od
el
's
tr
ai
nin
g
a
nd
vali
dation
acc
ur
ac
y
c
hange
d
t
hro
ughout
man
y
epo
c
hs.
Both
c
urves
are
sho
wing
sig
ns
of
impro
veme
nt,
wh
ic
h
mea
ns
the
m
od
el
is
le
arn
i
ng
a
nd
bec
om
in
g
bette
r
at
pro
pe
rly
cl
assif
yi
ng
occ
urrenc
es.
The
model
a
voids
over
fitt
ing
by
sim
ultan
eousl
y
im
pro
vi
ng
t
he
trai
ning
a
nd
va
li
dation
accu
ra
ci
es
since
the
y
are
qu
it
e
nea
r
eac
h
oth
e
r.
T
his
unit
y
sho
w
s
that
the
m
od
el
can
gen
e
rali
ze
we
ll
to
ne
w
dat
a,
stre
ng
t
he
nin
g
it
s
abili
ty
to
forecast
i
nf
ect
io
us
dise
ase
outc
ome
s
unde
r
su
r
veill
ance.
F
igure
6
s
hows
the R
NN
m
o
del
's trainin
g
a
nd
validat
io
n
los
s
thr
oughout ma
ny traini
ng epo
chs.
The
model
im
pro
ves
it
s
fit
to
t
he
trai
ning
data
as
ti
me
goes
on,
as
se
e
n
by
the
decr
e
asi
ng
tre
nd
of
bo
t
h
c
urves,
w
hich
minimi
ze
s
it
s
loss
f
unct
ion
.
Since
bot
h
the
t
rainin
g
and
validat
io
n
l
os
s
c
urves
c
onve
r
ge,
ind
ic
at
in
g
a
c
on
sist
e
nt
dr
op,
the
m
odel
is
unli
kely
t
o
be
over
fitt
ing
t
he
trai
ning
da
ta
.
This
c
ongr
uen
ce
impro
ves
the
model'
s
acc
ura
cy
i
n
inf
ect
ious
disease
surve
il
la
nce
predict
i
on
by
s
howi
ng
that
it
is
su
cce
ssfu
ll
y
gen
e
rali
zi
ng
to
ne
w
data.
Th
e
RN
N
m
odel
dem
onstrat
e
d
strong
f
unct
io
nalit
y
in
the
s
ys
te
m
f
or
mon
it
or
in
g
infecti
ous
dise
ases.
T
he
mod
el
showe
d
st
rong
pre
dicti
on
sk
il
ls,
s
uccessfull
y
disti
nguis
hing
betwee
n
i
ll
ness
and
non
-
diseas
e
cases
with
a
n
overall
acc
uracy
of
94%
a
nd
a
vali
dation
accurac
y
of
91
%.
F
urt
he
rm
ore,
the
model
dem
ons
trat
ed
it
s
ca
pa
ci
ty
to
ge
ne
rali
ze
well
to
unknown
data
without
ov
e
rf
it
ti
ng
by
ac
hievi
ng
a
l
ow
trai
ning
lo
ss
of
0.4
4
a
nd
a
val
idati
on
loss
of
0.55.
T
hese
fin
dings
dem
onst
rate
ho
w
well
the
R
NN
can
predict
fu
t
ur
e il
lness
pat
te
rn
s a
nd h
el
p wit
h publi
c he
al
th inter
ven
ti
on d
eci
si
on
-
ma
king.
Figure
5.
Ev
ol
ution o
f
trai
ning a
nd v
al
idati
on acc
ur
ac
y
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
1
2
3
4
5
6
7
8
9
10
Tr
aining
Accu
r
ac
y
Va
lidat
ion Ac
curac
y
Accura
cy
Va
lue
Epochs
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
omp E
ng
IS
S
N:
20
88
-
8708
Deep l
ear
ning
for infect
io
us
di
sease
s
ur
vei
ll
an
ce
integ
ra
ti
ng
inter
net
of th
ing
s
…
(
Subra
manian
Sumith
ra
)
1183
Figure
6. Tre
nd
of trainin
g an
d vali
datio
n
lo
ss
4.
CONCL
US
I
O
N
Infecti
ous
disease
surveil
la
nc
e
sy
ste
ms
are
e
ssentia
l
to
m
on
it
or
a
nd
man
ag
e
the
s
pr
ea
d
of
infecti
ons.
RNNs
to
the
se
sy
ste
ms
are
a
huge
ste
p
f
orwa
rd
since
t
hey
i
mpro
ve
their
a
bili
ty
to
see
tre
nd
s
,
ma
ke
deci
sion
s
,
and
ide
ntify
e
arly
warni
ng
s
ign
s
.
RN
N
models
ca
n
detect
outb
reak
s
of
diseases
by
e
xa
mini
ng
c
orrel
at
ion
s
and
patte
r
ns
i
n
la
rg
e
dataset
s
that
inclu
de
he
al
th
metri
cs
a
nd
e
nv
i
ronme
nt
al
var
ia
bles.
T
he
ca
pacit
y
t
o
adap
t
and
le
a
rn
f
rom
seq
ue
ntial
data
al
lows
them
t
o
un
derst
and
i
ntricat
e
correla
ti
on
s
and
ge
ner
at
e
pr
eci
s
e
pr
e
dicti
on
s
.
T
his
e
na
bles
he
a
lt
hcar
e
a
ut
hor
it
ie
s
to
act
qu
ic
kly
an
d
ma
ke
ef
fecti
ve
use
of
res
ources
.
High
accurac
y
a
nd
low
loss
value
s
ha
ve
be
en
pro
du
ce
d
by
va
li
da
ti
ng
RN
N
models
agai
ns
t
re
al
-
w
or
ld
data,
wh
ic
h
sh
ows
that
the
y
a
re
reli
a
ble
a
nd
suc
cessf
ul
i
n
disease
m
onit
or
in
g.
With
th
e
co
ntinuo
us
i
mpro
veme
nt
of
RN
N
te
chnolo
gy,
in
f
ect
iou
s
disease
monit
or
i
ng
mi
gh
t
be
c
omplet
el
y
tran
sf
orme
d,
al
lo
wing
public
healt
h
ma
nag
e
rs
to
be
more
pro
act
ive
an
d
le
ss
reacti
ve
.
T
he
r
e
m
us
t
be
on
goin
g
resea
rch
and
in
vestme
nt
in
t
hese
sy
ste
ms
t
o
us
e R
NN
-
ba
se
d
s
urveil
la
nce
sy
ste
ms
to
c
hal
le
ng
e
global
he
al
th concer
ns
f
ully.
REFERE
NCE
S
[1]
P.
Ho
m
la
an
d
K.
Pu
ritat,
“
Data
tran
sfo
rm
atio
n
in
th
e
in
fectiou
s
d
iseas
e
su
rveillan
ce
sy
stem
for
th
e
p
u
b
lic
h
ealth
in
form
atio
n
m
an
ag
em
en
t,
”
in
2
0
2
2
Jo
in
t
Inter
n
a
tio
n
a
l
Co
n
feren
c
e
o
n
Dig
ita
l
Arts,
Med
ia
a
n
d
Tech
n
o
lo
g
y
with
ECTI
No
rth
ern
S
ectio
n
Co
n
feren
ce
o
n
El
ectrica
l,
Electro
n
i
cs,
Co
mp
u
ter
a
n
d
Teleco
mmu
n
ica
tio
n
s
Eng
in
eering
(E
CTI
D
AMT
and
NC
ON)
,
Jan
.
2
0
2
2
,
p
p
.
2
1
0
–
21
3
,
d
o
i:
1
0
.11
0
9
/ECTI
DA
MT
NCO
N5
3
7
3
1
.2
0
2
2
.9720
4
2
3
.
[2]
R.
Krish
n
a
Van
ak
am
a
m
id
i,
L.
Ra
m
alin
g
am
,
N
.
Ab
ira
m
i,
S
.
P
riyan
k
a,
C.
S.
Ku
m
ar
,
an
d
S.
Muru
g
an
,
“
Io
T
secu
rity
b
ased
o
n
m
achi
n
e
le
arnin
g
,
”
in
2
0
2
3
S
econ
d
Inter
n
a
tio
n
a
l
C
o
n
feren
ce
On
S
m
a
rt
Tech
n
o
lo
g
ies
For
S
m
a
rt
Na
tio
n
(Sma
rtTech
Co
n
)
,
Au
g
.
2
0
2
3
,
p
p
.
6
8
3
–
6
8
7
,
d
o
i: 10
.110
9
/Sma
rt
TechC
o
n
5
7
5
2
6
.2
0
2
3
.1
0
3
9
1
7
2
7
.
[3]
T.
R.
Sa
ravan
an
,
A.
R.
Rath
in
am
,
J.
Lenin
,
A.
Ko
m
ath
i,
B.
Bh
arathi,
an
d
S.
Muru
g
an
,
“
Rev
o
lu
tio
n
izin
g
clo
u
d
co
m
p
u
tin
g
:
ev
alu
atin
g
th
e
in
f
lu
en
ce
o
f
b
lo
ck
ch
a
in
an
d
co
n
sen
su
s
alg
o
rithms,
”
2
0
2
3
3
rd
Inter
n
a
tio
n
a
l
Co
n
feren
ce
o
n
S
ma
rt
Gen
era
tio
n
Co
mp
u
tin
g
,
Co
mmu
n
ica
tio
n
a
n
d
Netw
o
rkin
g
,
S
MART
GENC
O
N
,
2
0
2
3
,
d
o
i: 10
.1109
/SMARTGE
NC
ON6
0
7
5
5
.20
2
3
.1
0
4
4
2
0
0
8
.
[4]
C.
H.
C
.
Sek
h
ar
,
V.
V
,
K.
Vijay
ala
k
sh
m
i,
M.
B.
S
ah
aai,
A.
S.
Rao
,
an
d
S.
Muru
g
an
,
“
Clo
u
d
-
b
ased
water
ta
n
k
m
an
ag
em
en
t
a
n
d
co
n
trol
sy
stem,
”
in
2
0
2
3
S
eco
n
d
Inter
n
a
tio
n
a
l
Con
feren
ce
On
S
ma
rt
Tech
n
o
lo
g
ies
For
S
ma
rt
Na
tio
n
(Sma
rtTech
Co
n
)
,
Au
g
.
2
0
2
3
,
p
p
.
6
4
1
–
6
4
6
,
d
o
i: 10
.110
9
/Sma
rt
TechC
o
n
5
7
5
2
6
.2
0
2
3
.1
0
3
9
1
7
3
0
.
[5]
S.
Kh
ad
d
a
j
an
d
H.
Ch
rie
f,
“
Preve
n
tio
n
an
d
co
n
trol
o
f
e
m
ergin
g
in
f
ectio
u
s
d
iseas
es
in
h
u
m
an
p
o
p
u
lati
o
n
s,
”
in
2
0
2
0
1
9
th
Inter
n
a
tio
n
a
l
S
ym
p
o
siu
m
o
n
Distr
ib
u
ted
Co
mp
u
tin
g
a
n
d
App
lica
tio
n
s
fo
r
Bus
in
ess
E
n
g
in
eerin
g
a
n
d
S
cien
ce
(
DCAB
ES)
,
Oct.
2
0
2
0
,
p
p
.
3
3
6
–
3
3
9
,
d
o
i: 1
0
.1
1
0
9
/DCAB
ES50
7
3
2
.2
0
2
0
.0009
2
.
[6]
S.
Hu
,
D
.
Ch
en
,
an
d
X.
Ch
en
g
,
“
Res
earc
h
o
n
earl
y
warnin
g
in
d
ex
sy
stem
o
f
in
fecti
o
u
s
d
iseas
es,
”
in
2
0
2
1
Inter
n
a
tio
n
a
l
Co
n
feren
ce
o
n
Pub
lic
Ma
n
a
g
ement
a
n
d
Intellig
en
t
S
o
ciety
(
PMIS)
,
Feb
.
2
0
2
1
,
p
p
.
3
9
6
–
3
9
9
,
d
o
i:
1
0
.1109
/PMI
S5
2
7
4
2
.20
2
1
.00
0
9
6
.
[7]
Q.
Tan,
J.
L
iu
,
B.
Sh
i,
Y.
Liu,
an
d
X
.
-
N.
Zho
u
,
“
Pu
b
lic
h
ealth
su
rveillan
ce
with
in
co
m
p
lete
d
ata
–
sp
atio
-
tem
p
o
ral
im
p
u
tatio
n
for
in
fer
ring
in
fectiou
s
d
iseas
e
d
y
n
am
ic
s,
”
in
2
0
1
8
IE
E
E
Inter
n
a
tio
n
a
l
Conf
eren
ce
o
n
Hea
lth
ca
re
Info
rma
tics
(I
CHI)
,
Ju
n
.
2
0
1
8
,
p
p
.
2
5
5
–
2
6
4
,
d
o
i:
1
0
.11
0
9
/ICHI.
2
0
1
8
.00
0
3
6
.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
1
2
3
4
5
6
7
8
9
10
Train
ing Loss
Va
lidat
ion Lo
ss
Loss
Va
lue
Epochs
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
1
175
-
1
186
1184
[8]
J.
Gen
g
,
Y.
Li,
Z
.
Zhan
g
,
an
d
L.
Tao,
“
Sen
tin
el
n
o
d
es
id
en
tification
for
in
fectiou
s
d
iseas
e
su
rveillan
ce
o
n
tem
p
o
ral
so
cial
n
etwo
rks
,
”
in
IE
E
E/WIC/
ACM
Inter
n
a
tio
n
a
l
Co
n
feren
ce
o
n
Web
Intelli
g
en
ce
,
Oct.
2
0
1
9
,
p
p
.
4
9
3
–
4
9
9
,
d
o
i: 10
.1145
/3
3
5
0
5
4
6
.3360
7
3
9
.
[9]
E.
Go
th
ai,
P.
Nat
esan
,
R.
R
.
Rajal
ax
m
i,
T.
Vig
n
esh
,
K.
Srinith
y
,
an
d
T.
V.
Balaji,
“
Pr
ed
ictiv
e
an
aly
sis
in
d
eterm
in
in
g
th
e
d
iss
em
in
atio
n
o
f
i
n
fectiou
s
d
iseas
e
an
d
its
sev
erity,
”
in
2
0
2
1
5
th
Inter
n
a
tio
n
a
l
Co
n
feren
c
e
o
n
Co
mp
u
tin
g
Meth
o
d
o
lo
g
ies
a
n
d
Co
mmu
n
ica
tio
n
(
ICCM
C)
,
Ap
r
.
2
0
2
1
,
p
p
.
1
5
5
6
–
1
5
6
2
,
d
o
i: 10
.1109
/ICC
MC5
1
0
1
9
.2
0
2
1
.9
4
1
8
2
2
8
.
[10
]
S.
Ray
an
asu
k
h
a,
A.
So
m
b
o
o
n
k
aew,
an
d
S.
Su
m
ridd
et
ch
k
ajo
rn,
“
Op
tical
sen
so
r
-
b
ased
m
as
s
tem
p
e
ratu
re
scr
e
en
in
g
n
etwo
rk
for
in
fectiou
s
d
iseas
e
su
rveillan
ce,
”
in
2
0
2
2
Co
n
feren
ce
o
n
La
ser
s
a
n
d
Electro
-
Op
tics
Pacific
Rim
(CLEO
-
P
R)
,
Ju
l.
2
0
2
2
,
p
p
.
1
–
2
,
d
o
i: 10
.1109
/CLEO
-
PR6
2
3
3
8
.2022
.
1
0
4
3
2
3
2
7
.
[11
]
W
.
Mahik
u
l
et
a
l.
,
“
A
Co
m
p
arat
iv
e
stu
d
y
o
f
tim
e
s
eries
an
aly
sis
for
in
fectiou
s
d
iseas
e
trend
s
in
Thailan
d
,
”
in
2
0
2
4
I
EE
E
Inter
n
a
tio
n
a
l
C
o
n
feren
ce
o
n
Big
Da
ta
a
n
d
S
ma
rt
Co
mp
u
tin
g
(B
ig
Co
mp
)
,
Feb
.
2
0
2
4
,
p
p
.
4
0
1
–
4
0
5
,
d
o
i: 10
.1109
/Big
C
o
m
p
6
0
7
1
1
.20
2
4
.0
0
0
8
9
.
[12
]
Inay
atu
llo
h
an
d
S.
Ther
esia,
“
Ea
r
ly
warnin
g
sy
stem
for
in
fectiou
s
d
iseas
es,
”
in
2
0
1
5
9
th
Inter
n
a
tio
n
a
l
Co
n
feren
ce
o
n
Teleco
mmu
n
ica
tio
n
Sys
tems S
ervices
an
d
A
p
p
lica
tio
n
s (
TS
S
A)
,
No
v
.
2
0
1
5
,
p
p
.
1
–
5
,
d
o
i: 10
.1
1
0
9
/TSSA.
2
0
1
5
.7
4
4
0
4
3
5
.
[13
]
M.
Jo
y
an
d
M.
Krish
n
av
en
i,
“
A
review
o
f
ep
id
em
ic
su
rveillan
ce
sy
stems
for
in
f
ectio
u
s
d
iseas
es,
”
in
2
0
2
2
Inter
n
a
tio
n
a
l
Co
n
feren
ce
o
n
Co
mp
u
tin
g
,
Co
mmu
n
ica
tio
n
,
S
ecur
ity
a
n
d
Intellig
en
t
S
ystems
(I
C3
S
IS)
,
Ju
n
.
2
0
2
2
,
p
p
.
1
–
6
,
d
o
i: 10
.1109
/IC3
S
IS54
9
9
1
.2022
.988
5
2
9
1
.
[14
]
J.
P.
Fitch
,
“
Eng
i
n
eering
a
g
lo
b
al
r
esp
o
n
se
to
in
fectiou
s
d
iseas
es,
”
in
Pro
ceedin
g
s
o
f
th
e
I
EE
E
,
v
o
l.
1
0
3
,
n
o
.
2
,
p
p
.
2
6
3
–
2
7
2
,
Feb
.
2
0
1
5
,
d
o
i: 10
.
1
1
0
9
/JPR
OC.2
0
1
5
.23
8
9
1
4
6
.
[15
]
A.
Bas
h
ir
,
A
.
W
.
Malik,
A.
U.
Rah
m
an
,
S.
Iqb
al,
P
.
R.
Cleary
,
an
d
A
.
Ikram,
“
Medclo
u
d
:
clo
u
d
-
b
ased
d
iseas
e
su
rveillan
ce
a
n
d
in
form
atio
n
m
an
ag
em
en
t sy
stem
,
”
IE
EE
Acce
ss
,
vo
l.
8
,
p
p
.
8
1
2
7
1
–
8
1
2
8
2
,
2
0
2
0
,
d
o
i: 1
0
.11
0
9
/ACC
ESS.
2
0
2
0
.29
9
0
9
6
7
.
[16
]
W
.
C.
Ko
k
,
J.
La
b
ad
in
,
’Aisy
ah
Moh
am
m
ad
,
K
.
S.
W
o
n
g
,
an
d
Y.
L.
Ch
an
g
,
“
An
d
roid
-
b
ased
d
iseas
e
m
o
n
ito
ring
,
”
in
2
0
1
6
Inter
n
a
tio
n
a
l
C
o
n
feren
ce
o
n
Info
rma
tio
n
a
n
d
Co
mmu
n
ica
tio
n
Tech
n
o
lo
g
y
(I
CICTM
)
,
2
0
1
6
,
p
p
.
9
7
–
1
0
3
,
d
o
i: 10
.1109
/ICICTM
.20
1
6
.78
9
0
7
8
4
.
[17
]
C.
Gu
ev
ar
a
an
d
M.
S.
Pen
as,
“
Su
rveillan
ce
rou
tin
g
o
f
COV
ID
-
1
9
in
fection
sp
read
u
sin
g
an
in
tellig
en
t
in
fectiou
s
d
iseas
e
s
alg
o
rithm,
”
IE
EE
Acce
ss
,
v
o
l.
8
,
pp
.
2
0
1
9
2
5
–
2
0
1
9
3
6
,
2
0
2
0
,
d
o
i: 10
.1
1
0
9
/
ACC
ESS.
2
0
2
0
.30
3
6
3
4
7
.
[18
]
Y.
Tang
,
X
.
Hu
,
a
n
d
S.
Li
,
“
An
ea
rl
y
d
etectio
n
sy
stem
for
d
iseas
e
o
u
tb
r
eaks
b
ased
o
n
d
en
sity
en
trop
y
,
”
in
2
0
1
9
Inter
n
a
tio
n
a
l
Co
n
feren
ce
o
n
Hig
h
Perfo
rm
a
n
ce
Big
Da
ta
a
n
d
Intellig
en
t
S
ystems
(HP
BD&IS
)
,
May
2
0
1
9
,
p
p
.
6
8
–
7
2
,
d
o
i: 10
.1109
/HPB
DIS.
2
0
1
9
.87
3
5
4
5
4
.
[19
]
L.
Natr
ay
an
,
M.
D.
R
.
Ka
m
al,
K.
K.
Maniv
an
n
an
,
a
n
d
G.
Su
n
il,
“
Mac
h
in
e
learnin
g
an
d
d
ata
m
in
in
g
ap
p
roach
es
for
in
fec
tio
u
s
d
iseas
e
su
rveillan
ce
an
d
o
u
tb
reak
m
an
ag
em
en
t
in
h
ealth
care
,
”
in
2
0
2
4
Int
ern
a
tio
n
a
l
Co
n
feren
ce
o
n
Adv
a
n
cements
in
S
ma
rt,
S
ecur
e an
d
I
n
tellig
en
t Co
mp
u
tin
g
(
ASS
IC)
,
Jan
.
2
0
2
4
,
p
p
.
1
–
7
,
d
o
i: 10
.11
0
9
/ASSIC6
0
0
4
9
.20
2
4
.10
5
0
7
9
9
0
.
[20
]
H.
-
M.
Lu,
D.
Z
en
g
,
an
d
H.
Ch
en
,
“
Pros
p
ectiv
e
in
fecti
o
u
s
d
iseas
e
o
u
tb
reak
d
etectio
n
u
sin
g
m
arko
v
switch
in
g
m
o
d
els,
”
I
EE
E
Tra
n
sa
ctio
n
s o
n
K
n
o
wled
g
e and Da
ta
E
n
g
in
eerin
g
,
v
o
l
.
2
2
,
n
o
.
4
,
p
p
.
5
6
5
–
5
7
7
,
Ap
r.
20
1
0
,
d
o
i: 10
.11
0
9
/TKDE
.20
0
9
.115.
[21
]
M.
S.
Rah
m
an
,
N
.
T.
Sa
fa,
S
.
Su
ltan
a,
S
.
S
ala
m
,
A.
Ka
ram
eh
ic
-
Mur
ato
v
ic,
an
d
H.
J.
Ov
erg
aard,
“
Ro
le
o
f
arti
ficial
in
te
llig
en
ce
-
in
ternet
o
f
th
in
g
s
(AI
-
IoT)
b
ased
e
m
ergin
g
tech
n
o
lo
g
ies
in
th
e
p
u
b
lic
h
ealth
resp
o
n
se
to
in
fectiou
s
d
iseas
es
in
Ban
g
lad
esh,
”
Par
a
site Epid
emio
lo
g
y an
d
Co
n
tro
l
,
v
o
l.
1
8
,
2
0
2
2
,
d
o
i:
1
0
.10
1
6
/j.parep
i.20
2
2
.e0026
6
.
[22
]
Q.
Jian
g
et
a
l.
,
“
I
n
tellig
en
t
m
o
n
ito
ring
for
in
fectiou
s
d
iseas
es
with
fuzzy
sy
stems
an
d
ed
g
e
co
m
p
u
tin
g
:
a
su
r
v
ey
,
”
App
lied
S
o
ft
Co
mp
u
tin
g
,
v
o
l.
1
2
3
,
Ju
l.
2
0
2
2
,
d
o
i:
1
0
.10
1
6
/j.aso
c.20
2
2
.10
8
8
3
5
.
[23
]
S.
Ch
ah
al,
“
D
eep
lea
rnin
g
for
earl
y
d
etectio
n
o
f
d
iseas
e
o
u
tb
reaks,
”
I
n
tern
a
tio
n
a
l
Jo
u
r
n
a
l
o
f
S
cien
ce
a
n
d
Resear
ch
(I
JS
R
)
,
v
o
l.
1
1
,
n
o
.
1
1
,
p
p
.
1
4
8
9
–
1
9
9
5
,
No
v
.
2
0
2
2
,
d
o
i: 1
0
.21
2
7
5
/SR2
3
1
0
0
3
1
6
2
3
2
1
.
[24
]
N.
T.
Nd
eh
,
Y.
T
.
Tes
faldet,
J.
Bu
d
n
ard,
an
d
P
.
Ch
u
a
ich
aroen
,
“
The
se
co
n
d
ary
o
u
tco
m
e
o
f
p
u
b
lic
h
ealth
m
easu
res
a
m
id
st
th
e
C
OVID
-
1
9
p
an
d
em
ic
in
th
e
sp
read
o
f
o
th
er
resp
ir
ato
r
y
in
fectiou
s
d
iseas
es
in
Thailan
d
,
”
Tra
vel
Med
icin
e
a
n
d
Infectio
u
s
Disea
s
e
,
v
o
l.
4
8
,
p
.
1
0
2
3
4
8
,
Ju
l.
2
0
2
2
,
d
o
i: 10
.
1
0
1
6
/j.tm
aid
.20
2
2
.
1
0
2
3
4
8
.
[25
]
H.
N.
Sah
a,
R
.
R
o
y
,
an
d
S.
Ch
ak
rabo
rty,
“
Clo
u
d
‐assis
ted
IoT
sy
stem
f
o
r
ep
id
em
ic
d
isea
se
d
etectio
n
an
d
sp
read
m
o
n
ito
ring
,
”
in
S
ma
rt Hea
lth
ca
re S
ystem Desig
n
,
W
iley
,
2
0
2
1
,
p
p
.
8
7
–
1
1
4
.
[26
]
K.
Karthik
a,
S.
D
h
an
alak
sh
m
i,
S.
M
.
Murth
y
,
N
.
Mish
ra,
S.
Sasik
ala
,
an
d
S.
Muru
g
an
,
“
R
asp
b
err
y
p
i
-
en
ab
le
d
wear
ab
le
sen
so
r
s
for
p
erso
n
al
h
ealt
h
trackin
g
an
d
an
a
ly
sis
,
”
in
2
0
2
3
Int
ern
a
tio
n
a
l
Co
n
fer
en
ce
o
n
S
elf
S
u
staina
b
le
Artificia
l
Intellig
en
ce
S
ystems
(I
CS
S
AS)
,
Oct.
2
0
2
3
,
p
p
.
1
2
5
4
–
1
2
5
9
,
d
o
i: 10
.11
0
9
/ICS
SAS5
7
9
1
8
.2023
.1
0
3
3
1
9
0
9
.
[27
]
S.
Selv
a
rasu
,
K
.
B
ash
k
aran,
K.
Rad
h
ik
a,
S.
Vala
rmat
h
y
,
an
d
S.
Muru
g
an
,
“
IoT
-
en
ab
led
m
ed
i
catio
n
safety:
real
-
tim
e
tem
p
e
ratur
e
an
d
sto
rage
m
o
n
it
o
ring
for
en
h
an
ce
d
m
ed
icatio
n
q
u
al
ity
in
h
o
sp
itals,
”
in
2
0
2
3
2
nd
Inter
n
a
tio
n
a
l
Co
n
feren
c
e
o
n
Auto
ma
tio
n
,
Co
mp
u
tin
g
a
n
d
R
en
ewa
b
le Sys
tems (IC
ACRS)
,
Dec.
20
2
3
,
p
p
.
2
5
6
–
2
6
1
,
d
o
i: 10
.11
0
9
/ICACR
S5
8
5
7
9
.20
2
3
.10
4
0
5
2
1
2
.
[28
]
D.
Sre
ed
h
aran,
M
.
S.
Su
b
o
d
h
Raj,
S.
N.
Geo
rge
,
an
d
S.
Ash
o
k
,
“
A
No
v
el
co
u
g
h
d
etectio
n
al
g
o
rithm
for
COV
I
D
-
1
9
su
rveillan
ce
at
p
u
b
lic
p
laces,
”
in
2
0
2
1
8
th
Inte
rn
a
tio
n
a
l
Co
n
fere
n
ce
o
n
S
ma
rt
C
o
mp
u
tin
g
a
n
d
C
o
mm
u
n
ica
tio
n
s
(I
CS
CC
)
,
Ju
l.
2
0
2
1
,
p
p
.
1
1
9
–
1
2
3
,
d
o
i:
1
0
.11
0
9
/ICSCC
5
1
2
0
9
.2021
.9528
2
9
5
.
[29
]
M.
R.
Hu
ss
ein
,
A
.
Bin
Sh
am
s,
E.
H
.
Ap
u
,
K.
A
.
Al
Ma
m
u
n
,
an
d
M.
S.
Ra
h
m
an
,
“
Dig
ital
su
r
v
eillan
ce
sy
stems
f
o
r
tracing
COV
ID
-
1
9
: priv
acy an
d
secu
rity ch
allen
g
es with
r
ecom
m
en
d
at
io
n
s,
”
a
rXiv:20
0
7
.13
1
8
2
,
2
0
2
0
.
[30
]
S.
Ch
ak
rabo
rty
et
a
l.
,
“
An
A
I
-
b
a
sed
m
ed
ical
ch
at
b
o
t
m
o
d
el
for
in
fectiou
s
d
iseas
e
p
redictio
n
,
”
IE
E
E
Acce
ss
,
v
o
l.
1
0
,
p
p
.
1
2
8
4
6
9
–
1
2
8
4
8
3
,
2
0
2
2
,
d
o
i: 10
.1
1
0
9
/ACC
ESS.
2
0
2
2
.32
2
7
2
0
8
.
[31
]
Gu
rm
in
d
erKa
u
r
an
d
B.
P.
Sin
g
h
,
“
Ex
p
lo
ration
an
d
p
red
ictio
n
an
aly
sis
o
f
featured
d
ata
sets
o
f
in
fectiou
s
d
iseas
e,
”
Ann
a
ls
o
f
th
e
Roma
n
ia
n
So
ciety for
Cell B
io
lo
g
y
,
v
o
l.
2
5
,
n
o
.
6
,
p
p
.
1
2
7
4
7
–
1
2
7
6
7
,
2
0
2
1
.
[32
]
S.
Gra
m
p
u
roh
it
an
d
C.
Sag
arnal,
“
Di
seas
e
p
redicti
o
n
u
sin
g
m
achi
n
e
learnin
g
alg
o
rithms,
”
in
2
0
2
0
Inter
n
a
ti
o
n
a
l
Co
n
fere
n
ce
fo
r
Emerg
in
g
Techn
o
lo
g
y (
INCET)
,
Ju
n
.
2
0
2
0
,
p
p
.
1
–
7
,
d
o
i:
10
.11
0
9
/INCET
4
9
8
4
8
.2
0
2
0
.9
1
5
4
1
3
0
.
[33
]
A.
Bald
o
m
in
o
s,
A.
Pu
ello
,
H.
Og
u
l,
T
.
Asu
rog
lu
,
an
d
R.
Co
lo
m
o
-
Palacio
s,
“
Predictin
g
in
fe
cti
o
n
s
u
si
n
g
co
m
p
u
ta
tio
n
al
in
tellig
en
ce
–
a sy
stematic
r
ev
i
ew,
”
IE
EE
Acce
ss
,
vo
l.
8
,
p
p
.
3
1
0
8
3
–
3
1
1
0
2
,
2
0
2
0
,
d
o
i:
1
0
.11
0
9
/ACC
ESS.
2
0
2
0
.2
9
7
3
0
0
6
.
[34
]
S.
Ag
rebi
an
d
A.
Lar
b
i
,
“
Us
e
o
f
a
rtif
icial
in
tellig
en
c
e
in
in
fectiou
s
d
is
eases,
”
in
Artifici
a
l
Intellig
en
ce
i
n
Precisio
n
Hea
lth
,
Elsev
ier,
20
2
0
,
p
p
.
4
1
5
–
4
3
8
.
[35
]
O.
E.
San
tan
g
elo
,
V.
G
en
tile,
S.
Pizz
o
,
D.
Gio
rdan
o
,
an
d
F.
Ced
ron
e,
“
M
a
ch
in
e
learnin
g
an
d
p
redictio
n
o
f
in
fe
ctio
u
s
d
iseas
es:
a
sy
stematic
revie
w,
”
Ma
ch
in
e
L
ea
rn
in
g
a
n
d
Kn
o
wled
g
e
Extra
ctio
n
,
v
o
l.
5
,
n
o
.
1
,
p
p
.
1
7
5
–
1
9
8
,
Feb
.
2
0
2
3
,
d
o
i: 10
.3390
/m
ak
e5
0
1
0
0
1
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.