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.
8
, No
.
6
,
Decem
ber
201
8
, p
p.
5245
~
5252
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
8
i
6
.
pp
5245
-
52
52
5245
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Analysin
g Tuber
culosis
T
re
nd
s
in
South A
sia
K
um
ar Abhis
hek
,
M.
P.
Sin
gh
,
M
d. Sa
dik
Hus
s
ain
Depa
rt
m
ent
o
f
E
le
c
tri
c
al a
nd
Co
m
pute
r
Scie
n
ce
and
Eng
ine
e
ring
,
Nat
ional
Insti
tu
te
of
T
ec
hnolog
y
,
Indi
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Sep
5
, 2
01
7
Re
vised
Ju
l
2
8
,
201
8
Accepte
d
Aug
11
, 201
8
Tube
rcu
losis
(T
B)
has
bee
n
one
of
the
top
te
n
c
ause
s
of
dea
th
i
n
the
world.
As
per
the
W
orld
Hea
lt
h
Org
an
iz
a
ti
on
(W
HO
)
aro
und
1.
8
m
il
l
i
on
peopl
e
have
di
ed
du
e
t
o
tube
r
cul
osis
i
n
2015.
Th
is
pa
per
a
ims
to
inv
esti
gate
th
e
spati
al
and
tem
pora
l
var
iatio
ns
in
TB
inc
i
dent
in
South
As
ia
(India
,
Bangl
ad
esh,
Pa
kista
n,
Ma
ldi
v
e
s,
Nepa
l
,
and
Sri
-
La
nka)
.
As
i
a
had
b
ee
n
count
ed
for
th
e
la
rge
st
n
um
be
r
of
new
TB
ca
ses
in
2015.
The
pap
er
under
li
n
es
and
rel
a
te
s
the
re
la
t
io
nship
bet
wee
n
v
ari
ous
fea
tur
es
l
ike
gende
r
,
age
,
loc
a
ti
on
,
oc
cur
ren
c
e,
and
m
orta
lit
y
due
to
T
B
in
the
se
coun
t
rie
s
for
the
per
iod
1993
-
201
2.
Ke
yw
or
d:
Data w
ra
ng
li
ng
Re
gr
essi
on
Tu
ber
c
ulo
sis
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Ku
m
ar Ab
hishe
k
,
Dep
a
rt
m
ent o
f C
om
pu
te
r
Scie
nce a
nd
E
ng
i
ne
erin
g,
Nati
on
al
I
ns
ti
tute o
f
Tec
hnol
og
y,
Patna
Patna
-
8000
05, B
ihar
,
I
nd
ia
.
Em
a
il
:
ku
m
ar.ab
his
he
k@nit
p.ac
.in
1.
INTROD
U
CTION
Tu
ber
c
ulo
sis,
abbre
viate
d
as
TB,
is
an
in
fecti
ou
s
disea
se
cause
d
by
Ba
ct
eria
Mycob
act
e
rium
tub
e
rcu
l
os
is
(
MTB
).
T
he
m
os
t
c
omm
on
r
ou
te
of
tra
ns
m
issi
on
is
ai
r
born
e
or
dro
plet
infecti
on.Pe
ople
with
act
ive
Pu
lm
onary
TB
spread
infecti
ous
aer
os
ol
dro
plets
of
0.5
to
5.0
µ
m
in
dia
m
eter
w
hile
co
ughing
,
sn
eezi
ng,
sp
ea
king,
si
ng
i
ng,
or
s
pitt
ing.
E
ach
one
of
th
ese
dro
plets
m
ay
transm
it
the
disease
,
since
th
e
infecti
ous
dose
of tu
ber
c
ulosi
s is v
e
ry sm
al
l
(the
i
nh
al
at
io
n of fe
wer t
ha
n 1
0 bacte
ria m
ay
cause a
n
i
nf
ect
ion).
The
lu
ngs
are
us
ua
ll
y
aff
ect
ed
by
TB
(Pul
m
on
ary
TB)
but
oth
e
r
body
par
ts
s
uch
as
Lym
ph
node
s,
kidney
,
bone
,
b
rai
ns
(E
xtra
P
ulm
on
ary
TB)
can
al
so
be
a
ffec
te
d.
It
m
a
y
be
fatal
if
no
t
treat
ed
pro
p
erly
as
it
is
a
sti
g
m
at
izing
disease.
Peop
le
in
so
m
e
cou
nt
ries
sti
ll
be
li
eve
that
it
is
an
incur
a
ble
conditi
on.
It
is
ver
y
diff
ic
ult
to
trea
t
the
patie
nts
if
they
do
no
t
fol
low
the
re
gim
en
ti
gh
tl
y.
Th
e
re
are
on
ly
fe
w
antibioti
cs
tha
t
can
be
us
e
d
a
gains
t
it
and
resist
ance
em
erg
es
r
eadil
y.
The
va
cci
nes
of
TB
are
not
com
plete
ly
eff
ect
ive.
MTB
sp
rea
ds
by
res
pirato
ry
dro
ple
ts
and
dies
in
s
un
li
ght.
T
rans
m
issi
on
does
not
ha
pp
e
n
im
m
ediat
el
y,
bu
t
re
qu
i
res
so
m
e
pr
ol
onge
d
e
xposure
.
W
e
m
a
y
infer
th
at
it
is
associa
te
d
w
it
h
po
or
dark
a
nd
ding
hy
li
ving
c
ondi
ti
on
s.
The per
son
w
ho is at
higher
r
i
sk
of TB incl
udes:
A pers
on
with
weak i
m
m
un
e syst
e
m
.
A pers
on
with
poor
nu
trit
io
na
l st
at
us
.
Drug a
dd
ic
ts.
A pers
on s
uffe
rin
g
f
r
om
H
IV
infecti
on
or
dia
betes.
TB
is
on
e
of
t
he
m
ajo
r
fa
ct
ors
of
death
a
nd
disabili
ty
in
t
he
w
or
l
d.
Acc
ordin
g
to
a
repor
t
by
WH
O
,
there
a
re
a
ppr
oxim
a
te
ly
9.
6
m
il
li
on
cases
of
TB
,
in
w
hich
2.2
m
illi
on
cas
es
are
i
n
India
on
ly
.
The
TB
cases
in
I
nd
ia
acc
ou
nts
f
or
$3
40
bi
ll
ion
to
the
I
ndia
n
ec
onom
y.
TB
is
al
so
a
l
e
adin
g
kill
er
of
pe
ople
af
fected
by
HIV
resu
lt
in
g
for
a
bout
35
%
of
t
he
deaths
.
The
m
ai
n
caus
es
of
TB
e
pid
e
m
ic
are
il
li
te
ra
cy
,
i
m
pr
oper
r
eso
ur
ce
distrib
ution, i
m
pr
op
er
d
ie
ts
and lac
k o
f pro
per m
edicat
ion
.
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.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5245
-
52
52
5246
Accor
ding
to
a
repor
t,
ab
out
49
m
il
li
on
li
ves
of
TB
p
at
ie
nt
s
wer
e
save
d
be
tween
2000
-
2015
th
r
ough
TB
diag
no
sis
a
nd
treat
m
ent.
The
ob
j
ect
ive o
f
this pap
e
r
is
to
analy
ze
TB
trend
i
n
South
Asia
beca
us
e
this
can
help
by
pro
vid
i
ng pr
op
e
r kno
wled
ge of
the c
ases, thei
r
rea
s
on
s
, ge
ographi
cal
d
ist
rib
utio
n, et
c..
2.
RE
LATE
D
W
ORK
Ma
ny
sim
il
ar
works
ha
ve
be
en
done
on
T
uberc
ulo
sis
pre
dicti
on
.
Ma
ny
sch
olars
a
dopt
ed
diff
e
ren
t
m
et
ho
dolo
gies.
Noo
dch
a
nath
et
.
a
l
.
m
od
el
e
d
the
incide
nc
e
of
TB
cases
in
T
haila
nd
[
1]
.
T
heir
ai
m
was
t
o
desig
n
a
m
od
el
wh
ic
h
a
naly
zes
the
trend,
seaso
nal
an
d
geogr
a
phic
al
incident
of
T
B
cases
in
southe
rn
Thail
an
d
f
ro
m
1999
to
2004.
They
ha
ve
use
d
Ne
gative
B
ino
m
ia
l
Distri
bu
ti
on
for
gender,
age
,
locat
i
on
a
nd
for
oth
e
r
var
ia
ble
L
og
-
li
near
Dis
trib
utio
n
was
us
e
d.
A
fter
getti
ng
t
he
di
stribu
ti
on
t
he
y
hav
e
a
pp
li
ed
Linea
r
regressin
g
t
o
be
st
fit
acco
rd
i
ng
to
distri
bu
ti
on.
Af
te
r
the
an
al
ysi
s,
they
rea
li
zed
that
both
m
al
e
and
f
em
a
le
has
the
sam
e
risk
of
TB
hav
i
ng
a
ge
le
ss
than
25.
W
it
h
the
i
ncr
e
ase
in
the
a
ge
,
in
both
m
al
es
and
fem
al
es
risk
al
s
o
increases
.
The
r
e
is
no
t
any
se
aso
nal
trend
in
the
TB
disease
cases.
Ther
e
are
long
-
te
rm
seaso
nal
chang
es
in
TB
cases
from
1999
to
2004.
The
Ge
ogra
phic
al
locat
io
ns
al
so
a
ff
e
ct
T
B
cases.
I
n
upper
weste
rn
an
d
lo
w
e
r
so
ut
hern t
her
e
is a hi
gh r
is
k o
f
TB case
s.
Sam
pu
rn
a
et
.
al
.
al
so
worked
on
a
sim
i
la
r
proj
ect
[
2]
.
They
hav
e
m
od
el
ed
TB
incident
in
Ne
pal.
I
n
Nep
al
,
t
he
TB
case
rate
is
ve
r
y
hig
h,
s
o
thei
r
ob
j
ect
ive
wa
s
to
m
od
el
incident
of
TB
in
Nep
al
from
2003
t
o
2008.
They
use
d
gende
r
a
nd
locat
ion
as
a
pa
ram
et
er
to
an
al
yz
e
the
TB
c
ases.
T
o
def
i
ne
the
distrib
utio
n,
t
hey
us
e
d
the
Neg
at
ive
Bi
no
m
ia
l
t
echn
i
qu
e
.
Af
te
r
the
distrib
ution,
they
us
ed
li
near
regressio
n
to
get
the
tre
nd
of
TB
cases
and
pr
e
dict
the
resu
lt
with
two
m
u
lt
ipli
cat
ive
com
po
nen
ts
as
predict
ors.
T
hey
fo
un
d
that
there
ar
e
m
ajo
r
dr
op
s
in
TB
cases
du
ri
ng
these
fi
ve
ye
ars
in
Nep
al
.
Av
e
rag
e
TB
cases
in
m
al
e
wer
e
1.3
1
pe
r
1000
popula
ti
on
,
whic
h
is
ver
y
le
ss
an
d
i
n
the
cas
e
of
a
fem
al
e,
it
is
1.81.
In
N
epal
re
gion
p
la
y
a
vital
r
ole
i
n
th
e
case
of
TB
.
I
n
Terr
ai
n
re
gion,
it
is
hig
her
fo
l
lowe
d
by
Hill
and
M
ountain
reg
i
on.
The
re
i
s
a
decr
eas
e
in
tren
d,
bu
t
sti
ll
,
a
total
nu
m
ber
of
T
B
cases
is
ver
y
hig
h.
T
he
Hi
gh
e
r
rate
is
in
Terr
ai
n
reg
i
on
and
urban
a
rea
.
They
analy
ze TB
cas
e on a l
ong
te
r
m
b
asi
s.
Sam
pu
rn
a
et
.
al
.
a
naly
zed
t
he
te
m
po
ral
an
d
s
patia
l
va
riat
ion
of
TB
cas
es
in
Ne
pal
[
3]
.
Data
wa
s
colle
ct
ed
f
ro
m
2003
to
20
10.
Data
is
m
od
el
ed
by
us
in
g
ge
nd
e
r,
a
ge,
loc
at
ion
,
ye
ar
us
i
ng
li
near
regre
ssion
m
od
el
with
log
trans
form
a
tio
n
of
the
rate
of
the
pa
ram
eter
.
They
rem
ov
ed
s
om
e
ou
tl
i
er
to
get
a
good
fit
of
the
data
i
n
the
m
od
el
.
HIV
w
as
al
so
c
onside
red
f
or
getti
ng
the
va
riat
ion
in
the
m
od
el
.
It
is
seen
that
HIV
case
le
ads
the i
ncr
e
ase in TB cas
e
s.
T
he
rate
of T
B case
s v
a
ries
hi
gher i
n
Te
rr
a
in r
e
gion a
nd T
B case
s, variat
ion o
f
locat
ion
an
d
ye
ars.
A
work
on
da
ta
acqu
isi
ti
on
and
analy
sis
of
s
olar
e
nergy
ge
ner
at
io
n
w
as
car
ried
.
Th
e
pa
ram
et
ers
consi
der
e
d
f
or
analy
sis
we
re
nam
ely
(1)
a
ve
rag
e
(
2)
m
axi
m
u
m
and
(
3)
total
am
ou
nt
of
el
ect
ri
ci
ty
generate
d
on d
ai
ly
and m
on
t
hly basis.
P
red
ic
ti
on
was p
erfor
m
ed
after
analy
sis usi
ng
ANN
m
od
el
[10]
.
The
ab
ov
e
a
na
ly
sis
is
fo
cuse
d
on
a
par
ti
cul
ar
country
wit
h
locat
ion,
age
,
sex
as
par
a
m
et
ers.
They
al
so
co
ns
ide
r
the
seaso
nal
an
d
long
-
te
rm
tr
end.
The
pro
pose
d
w
ork
ana
ly
ses
TB
trends
in
al
l
cou
ntri
es
of
South
Asia.
T
he
par
am
et
ers
con
si
der
e
d
a
re
Ag
e
,
Ge
nder
,
Locati
on
an
d
HIV
Ca
ses
as
HIV
re
duces
a
per
s
on’s
i
m
m
un
it
y
resul
ti
ng
in
an
inc
reased
c
han
ce
of
TB
.
T
o
fill
m
issi
ng
valu
es
ge
ne
ral
sta
ti
sti
cs
li
ke
m
e
a
n
a
nd
m
edian
are
not
u
se
d
i
ns
te
ad
a
m
achine lear
nin
g m
od
el
is tra
ined
t
o pr
e
dict
the m
issi
ng
v
al
ues.
3.
DA
T
A ANAL
YS
IS
The
data
for
a
naly
sis
has
be
en
obta
ine
d
f
r
om
a
W
H
O
re
port
on
TB
f
or
al
l
the
countri
es.
The
ra
w
data
of
six S
ou
th
Asia
n
co
unt
ries
viz.
India, Bang
la
desh,
S
r
i
Lank
a
,
Pa
kistan,
Ma
ldi
ves,
a
nd
N
e
pal
ha
ve
bee
n
scru
ti
nize
d.
S
om
e
data
wer
e
pr
e
dicte
d
us
in
g
the
m
et
ho
d
of
Li
near
Re
gressi
on.
T
he
ra
w
data
was
pr
ocesse
d
thr
ough the
foll
ow
in
g p
ro
ces
s
es.
a.
Data cl
eani
ng
b.
Data vali
da
ti
on
c.
Data w
ra
ng
li
ng
d.
Line
ar
re
gr
essi
on
The
whole
data
set
was
cl
ass
ifie
d
m
et
ho
dolog
ic
al
ly
to
determ
ine
the
country
wise
ne
w
pu
lm
on
a
r
y
cases ba
sed
on se
x, locat
io
n,
a
nd HIV
posit
iv
e cases.
4.
PROBLE
M
F
ORMUL
ATI
ON
The
dataset
av
ai
la
ble
hav
e
s
om
e
m
issi
ng
values
,
ne
gative
value
s,
an
d
O
utli
ers.
S
o
the
dataset
is
cl
eaned
f
ro
m
su
ch
un
wan
te
d
values
.
T
he
li
ne
ar
m
od
el
is
prepa
red
a
nd
an
al
ysi
s
is
done
by
R
la
ng
ua
ge
.
T
he
work is
done st
ep by st
ep
whi
ch follo
ws:
-
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
An
alysing T
uberculosis
tren
ds i
n S
ou
t
h
Asi
a
(
Kumar A
bhis
hek
)
5247
a.
Find o
ut
neg
at
i
ve value a
nd
fill
the
m
w
it
h nul
l values.
b.
Sp
li
t datase
t i
nt
o
an
avail
a
ble
dataset
and m
i
ssing dataset
.
c.
Pr
e
par
e a
li
nea
r
m
od
el
d.
Apply o
utli
er re
m
ov
al
alg
or
it
hm
e.
Pr
e
dict m
issi
ng
values
f.
Fil
te
r
releva
nt
data
g.
An
al
yz
e
datase
t by R
qu
e
ry
So
m
e
al
go
rith
m
s
are
app
li
ed
to
the
dataset
to
analy
ze
and
ob
ta
in
a
m
eaningf
ul
resu
lt
.
Fi
rst
of
al
l
the
neg
at
ive
val
ue
is
found
ou
t
a
nd
re
placed
by
nu
ll
val
ue.
T
he
reafter
dataset
is
sp
li
t
into
tw
o
pa
rts.
T
he
fir
st
par
t
is
correct
data
and
the
seco
nd
pa
rt
is
m
issi
ng
data.
Al
gorith
m
1
is
us
e
d
f
or
re
placi
ng
m
iss
ing
valu
es
with
nu
ll
value
Algori
th
m
1
:
f
il
lNegati
veV
al
ue
Inpu
t:
dataset
Out
p
ut:
datas
et
w
it
hout
ne
ga
ti
ve
val
ue
FO
R
r
ow IN
d
at
aset
:
FO
R c
olu
m
n
I
N
r
ow:
IF
(
dataset
[c
olum
n]<0)
:
Fr
om
the
co
rrec
t
dataset
,
a
m
od
el
is
desig
ned
so
that
th
e
m
issi
ng
val
ue
s
can
be
pr
e
di
ct
ed.
S
om
e
relat
ed
value
s
wh
ic
h
wer
e
av
ai
la
ble
in
m
iss
ing
valu
e
r
ow
are
passe
d
as
an
in
put
to
the
prepa
red
m
odel
and
we
get
data
for
m
issi
ng
valu
es.
This
predic
te
d
value
is
not
act
ual
value
,
bu
t
we
try
to
ob
ta
in
val
ue
s
with
m
axi
m
u
m
eff
ic
ie
ncy
.
T
o
te
st
the
acc
ur
acy
of
t
he
m
od
el
f
r
om
cor
re
ct
ed
dataset
,
tw
o
s
ub
-
dataset
s
are
fou
nd.
On
e
is
trai
ni
ng
data
a
nd
the
s
econd
one
is
te
st
data.
T
rain
da
ta
is
us
e
d
to
t
rain
t
he
m
od
el
and
te
st
data
is
use
d
to
te
st
the
acc
ur
acy
of
the
m
od
el
.
S
om
e
var
ia
ti
on
s
are
do
ne
i
n
the
pa
ra
m
et
ers
of
the
m
od
el
,
so
t
hat
a
good
datam
od
el
can be
ob
ta
ine
d. Al
og
i
rtm
2
is use
d for this
purp
os
e
Algori
th
m
2 :
trai
nModel
Inpu
t:
dataset
Out
p
ut:
li
nea
r data
m
od
el
data=rea
d_cs
v(dataset
)
re
g=li
near_m
od
el
.
Linea
rRegr
essi
on(
)
RETUR
N reg
The
pr
e
pa
red
m
od
el
is
a
Linear
Re
gr
essio
n
m
od
el
.
T
his
is
the
m
od
el
wh
ic
h
us
es
som
e
existi
ng
featur
e
s
by
w
hi
ch
a
li
nea
r
eq
uation
is
obta
ined
re
pr
ese
nti
ng
the
pr
e
dicte
d
value.
The
equ
at
io
n
is
ob
ta
ined
with
le
ss
Re
sidu
al
Square
S
um
(RSS)
wh
ic
h
is
su
m
of
the
sq
ua
re
of
the
diff
e
re
nce
bet
ween
act
ual
va
lue
and
pr
e
dicte
d
valu
e.RSS
is
us
e
d
to
pr
e
dict
co
ntinuo
us
value
s.
I
n
t
his
pr
oject
,
al
l
values
are
c
onti
nuou
s
rea
l
nu
m
ber
s.
T
hu
s
, th
e li
ne
ar m
od
el
is reall
y
helpful.
Ther
e
a
re
so
m
e
ou
tl
ie
r
val
ues
in
the
dataset
,
wh
ic
h
can
no
t
be
co
rr
ect
e
d
by
li
near
re
gress
ion
.
Ou
tl
ie
r
pro
du
ces
m
or
e
RSS
value
,
due
to
wh
ic
h,
li
ne
ar
re
gressi
on
i
s
away
from
a
good
fit.
S
o
a
n
al
gorithm
is
app
li
ed
to
increase
t
he
accuracy
of
th
e
m
od
el
.
Algo
r
it
h
m
3
is
app
li
ed
to
a
trai
ni
ng
dataset
and
ea
ch
tim
e
so
m
e
ou
tl
ie
r
is rem
ov
ed
an
d t
he
rem
ai
nin
g datase
t m
od
el
i
s r
e
-
trai
ned. T
hi
s p
r
ocess
is re
peated
un
ti
l m
axim
u
m
accur
acy
is
ob
ta
ine
d .
Algori
th
m
3:
r
e
m
ov
eO
utli
er
Inpu
t:
predi
ct
ion
, feat
ure,
targ
et
Out
p
ut:
ou
tl
ie
r
rem
ov
e
d data
set
te
st=
np
.a
rr
ay
(
[
]
)
x=np.
a
ppen
d(t
est
,f
eat
ures)
x=x[:9
0]
y=
np
.a
ppen
d(t
est
,targ
et
)
y=
y[:
90
]
z=np.ap
pe
nd(t
est
,p
re
dicti
ons)
z=z[:
90]
tem
p=n
p.
a
rray
([x,
y,a
bs
(z
-
y)]
)
resu
lt
=t
em
p.
transpose
()
resu
lt
.s
or
t
()
le
ng
th=l
e
n(res
ult)
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.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5245
-
52
52
5248
sta
rt=
m
at
h.
flo
or(len
gth*0.1
)
cl
eaned_da
ta
=
r
es
ult[start
+1:
]
RETURN
clea
ned_
data
Fr
om
ab
ov
e
proce
ss,
a
bette
r
accurate
m
od
el
is
obta
ine
d.
N
ow,
al
l
m
issi
ng
data
a
re
pr
e
di
ct
ed
us
i
ng
al
gorithm
4.
Algori
th
m
4 :
pr
e
dicti
on
Inpu
t:
dataset
,input
Out
p
ut:
predi
ct
ed
val
ue
trai
n,
te
st=
trai
n_te
st_s
plit
(d
at
a
set
,test
=3)
l
m
= li
near
Mode
l(dataset
)
WHILE
(accu
r
acy
<
exp
ect
e
d_accuracy)
DO
cl
ean_data
=re
m
ov
eOu
tl
ie
r(
l
m
.p
red
ic
t(te
st[
featur
e
s]
))
l
m
.f
it
(clean_da
ta
[f
eat
ures]
,clea
n_data[ta
rg
et
]
)
RETURN
lm
.p
red
ic
t(cl
ea
n_da
ta
[input]
)
The
res
ult
sho
ws
that
the
pr
e
dicti
on
of
m
iss
ing
values
us
i
ng
t
he
m
achine
le
arn
in
g
m
od
el
ar
e
m
or
e
pr
eci
se
d
th
an
the
gen
e
ral
sta
t
ist
ic
al
m
od
el
.
Af
te
r
getti
ng
t
he
m
issi
ng
va
lues,
the
datase
t
beco
m
es
co
m
ple
te
and
acc
ur
at
e.
Since
analy
sis
is
done
only
for
South
Asian
Countries
with
so
m
e
featur
es
(Y
ear
,
A
ge,
R
egio
n,
and
Ne
w
TB
Ca
ses
et
c.),
so
from
th
e
who
le
dataset
re
quired
dataset
is
ob
ta
ine
d.
Th
en
al
l
the
data
se
ts
are
analy
zed t
o gr
a
b
the
in
form
at
i
on u
si
ng R La
ngua
ge.
Fo
r
di
ff
e
ren
t
c
ountries,
we
ge
t
a
dif
fer
e
nt
li
near
eq
uatio
n
and
acco
r
ding
to
that
li
nea
r
e
qu
at
io
n,
th
e
m
issi
ng
v
al
ues
are pre
dicte
d.
So
m
e o
f
t
hem
are m
entioned
as foll
ows
-
Fo
r
Ban
glade
s
h,
=
5683
−
1
1
3
1
7
4
60
wh
e
re y is
ne
w
sm
ear
-
po
sit
ive
case a
nd q is t
he
ye
ar
for w
hi
ch values
are g
oing to
b
e
pre
di
ct
ed.
Sim
il
arl
y,
Fo
r
In
dia,
=
25749
−
5
1
1
3
9
722
Fo
r
Mal
div
e
s,
=
−
5
+
10091
Fo
r
Ne
pal,
=
−
30
(
−
2010
)
2
+
1556
To
c
onve
rt the
above
e
quat
io
n
into t
he
li
nea
r e
qu
at
io
n, le
t
(
−
2010
)
2
=
T
so
,
=
−
30
+
1556
Fo
r
Pa
kistan,
=
(
698
(
−
1998
)
2
)
+
2567
le
t
(
−
1998
)
2
=
so
,
=
698
+
2567
Fo
r
S
ri La
nk
a
,
=
−
5
(
−
2009
)
2
+
4764
le
t
(
−
2
009
)
2
=
so
,
=
−
5
+
4764
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
An
alysing T
uberculosis
tren
ds i
n S
ou
t
h
Asi
a
(
Kumar A
bhis
hek
)
5249
Si
m
il
arly
,
afte
r
fin
ding
a
li
ne
ar
eq
uatio
n
f
or
al
l
va
riable,
m
issi
ng
value
s
are
pre
dicte
d.
Now,
on
these
data
set
s
analy
sis
w
ork
i
s
done
an
d
s
om
e
facts
ar
e
f
ound
out
w
hich
are
ex
plaine
d
i
n
the
res
ult
a
na
ly
sis
sect
ion
.
5.
RESU
LT
S
A
ND AN
ALYSIS
The
analy
sis
is
fo
cuse
d
on
TB
data
for
South
As
ia
n
Countries
(m
a
inly
In
di
a,
Pakistan
,
Nep
al
,Ba
ngla
de
sh
,Mal
di
ves)
from
W
H
O
for
a
pe
rio
d
of
1993
-
20
12.
The
analy
sis
show
s
that
I
nd
ia
ha
s
th
e
highest
cases
r
egiste
red
for
ne
w
sm
ear
-
po
si
ti
ve
TB
wh
ere
as
the
Ma
ldives
has
the
le
ast
cases
rep
ort
e
d
f
or
new
sm
ear
-
po
sit
ive
TB(
de
pi
ct
ed
in
Fig
ure1).
C
on
si
der
i
ng
gender
wise
data
f
or
ne
w
sm
ear
-
posit
ive
T
B
cases,
m
al
e
be
tween
the
a
ge
gro
up
of
35
t
o
44
a
re
m
os
t
t
o
ha
ve
rec
orde
d
f
or
new
case
s
of
TB
as
s
h
own
i
n
Figure
2
.
Fem
al
es
between
a
ge
gro
up
15
-
24
ye
ars
rec
ord
ed
the
hi
gh
e
st
cases
for
ne
w
sm
ear
-
po
sit
ive
TB
as
sh
ow
n
in
Fi
gur
e 3
.
Figure
1. Ne
w sm
ear
-
po
sit
ive
case co
untry
wi
se
Figure
2. Ne
w sm
ear
-
po
sit
ive
case co
untry
wi
se f
or
m
al
e age
wise
Figure.
3.
New sm
ear
-
po
sit
ive
case c
ountry
wise fo
r
fem
ale
age
W
it
h
resp
ect
t
o
Fi
gure
1,
Fig
ur
e
2
a
nd
Fi
gure
3
I
ndia
has
the
highest
num
ber
of
ne
w
s
m
ear
-
posit
iv
e
TB
cases
colle
ct
ively
as
well
as
gen
de
r
wis
e
al
so
India
reco
r
de
d
the
highest
cases,
f
ollow
e
d
by
Ba
ng
la
desh,
Paks
ta
n,
Nepal
, S
ri La
nk
a
. Mal
div
es
has
the
least
n
um
ber
o
f new sm
ear
-
po
sit
ive TB ca
ses co
ll
ect
ively
f
or th
e
whole
popula
ti
on as
well
as
ge
nd
e
r
-
wise.
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.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5245
-
52
52
5250
Figure
4
dep
ic
ts
that
there
ha
s
been
a
co
ns
t
ant
increase
in
th
e
nu
m
ber
of
patie
nts
reg
ist
ered
under
new
sm
ear
-
po
sit
ive
TB
cases
ever
y
ye
ar
desp
it
e
the
e
ffor
ts
of
WHO
an
d
res
pect
ive
co
un
t
ries
towa
r
d
eff
ect
ive
im
ple
m
entat
ion
of
t
he
Sto
p
TB
Stra
te
gy.
Figure
4
.
Ne
w sm
ear
-
po
sit
ive
case y
ear
wise
Accor
ding
to
Figur
e
5,
I
ndia
co
un
ts
f
or
t
he
highest
num
ber
Re
turn
Re
la
ps
e
cases
f
ollow
e
d
by
Ba
ng
la
desh,
P
akista
n,
Sr
i
La
nk
a
,
Ne
pal.
T
he
return
Re
plase
case
ind
ic
at
es
tho
se
patie
nt
s
who
are
i
de
ntifie
d
as
sm
ear
-
po
sit
ive
ta
b
a
nd
t
he
y
hav
e
m
issed
their
treat
m
ent
reg
im
e.
The
P
at
ie
nts
identifi
ed
as
Re
tu
rn
r
el
apse
cases ha
ve
a
h
i
gh ch
a
nce
of
MDR (M
ulti
-
D
rug
Re
sist
ant
) TB
which i
n re
cent ye
ar
has b
een in
c
reasi
ng.
Figure
5. Re
tur
n rela
ps
e ca
se
country
wise
Accor
ding
to
Figure
6,
it
can
be
seen
that
after
2011,
Re
tur
n
relapse
c
a
ses
sta
rt
decr
e
asi
ng
due
to
awar
e
ness
ab
out
m
edical
cures
f
or
TB.
WH
O
e
xecu
te
s
lots
of
a
war
e
ness
pro
gr
am
s
wh
ic
h
pays
po
sit
ive
eff
ect
on Ret
urn rel
a
ps
e
case
s.
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
An
alysing T
uberculosis
tren
ds i
n S
ou
t
h
Asi
a
(
Kumar A
bhis
hek
)
5251
Figure
6. Re
tur
n rela
ps
e ca
se
ye
ar w
ise
People
suffe
ri
ng
f
ro
m
HIV
hav
e
26
to
31
tim
es
gr
eat
er
chan
ce
s
of
TB
than
pe
op
le
without
HIV.
W
it
h
res
pect
t
o
WHO
re
port
in
2014
,9.6
m
illi
on
ne
w
ca
ses
of
TB
we
r
e
reg
ist
ere
d
out
of
wh
ic
h
1.2
m
illi
on
wer
e
people
wi
th
HIV.
T
he
pe
op
le
s
uffer
i
ng
from
HI
V
ha
ve
high
cha
nces
of
getti
ng
TB
if
they
are
in
co
nta
ct
with
TB
suff
e
rin
g
people
be
cause
of
thei
r
reduce
d
im
m
u
nity
.
The
TB
in
H
I
V
people
can
be
c
ur
e
d
with
pro
per
treat
m
e
nt
re
gim
e
and
su
r
veill
ance.
W
it
h
res
pect
t
o
Fig
ure
7
Ind
ia
has
m
or
e
than
50
per
ce
nt
of
HIV
reg
ist
ere
d
pati
ents
are
af
fected
by
TB.
Ac
cordin
g
to
Fi
gure
8,
the
inc
r
ease
in
the
pe
rcen
ta
ge
of
cases
of
TB/
HIV is alm
os
t c
on
sta
nt af
t
er
2009.
Figure
7
.
H
IV
and TB case
coun
t
ry w
ise
Figure
8. H
IV
and TB case
yea
r wise
6.
CONCL
US
I
O
N
AND
F
UT
U
RE EN
HAN
C
EMENT
The
pap
e
r
s
hows
var
io
us
va
riat
ion
s
i
n
the
TB
cases
ba
s
ed
on
t
he
se
x,
locat
ion,
ye
ar
,
an
d
HIV
-
aff
ect
ed
case
s.
It
can
be
us
ed
as
an
ai
d
f
or
r
e
con
ci
li
at
ion
a
nd
a
rethin
ki
ng
of
the
a
ppr
oac
h
ad
op
te
d
earli
er
an
d
the
furthe
r
im
p
rovem
ent
needed
to
accom
plish
the
goal
of
1case
pe
r
m
i
ll
i
on
per
ye
ar
of
W
H
O
by
2050
.
The
pr
ese
nted va
riat
ion
can help i
n
f
ocusi
ng and
taking
a tar
get
ed
ap
proac
h
f
or the b
a
sem
ent
of
the
w
or
st a
f
fected
reg
i
on.
This
pa
per
pro
vid
es
an
analy
sis
of
data
relat
ed
to
So
ut
h
Asia
w
it
h
a
lim
it
ed
nu
m
be
r
of
par
a
m
et
ers
wh
ic
h
ha
ve
be
en
w
hich
ha
ve
been
ta
ke
n
into
con
side
rati
on.
In
fu
t
ur
e d
at
as
et
can
be
analy
zed
for
al
l
countries
of the
w
or
ld
w
i
th s
om
e
m
or
e p
aram
et
ers.
T
he
sam
e w
ork
c
an be
done f
or
analy
sis o
f oth
er
diseases.
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.
8
, N
o.
6
,
Dece
m
ber
201
8
:
5245
-
52
52
5252
REFERE
NCE
S
[1]
Kongchou
y
N,
Kakc
hapati
S,
Choonpradub
C.
“
Mode
li
ng
the
inc
ide
n
ce
of
tu
ber
cul
osis
in
southern
Thaila
nd
”
.
Southe
ast
Asian Journal
of
Tr
opic
al
M
edi
c
ine an
d
Public
H
eal
th
.
1
;
41
(3)
:574
,
20
10
.
[2]
Kakc
hapati
S,
Choonpradub
C
,
L
im
A.
“
Spat
ia
l
and
te
m
pora
l
var
ia
t
ions
in
tube
rcu
losis
incidence,
N
epa
l
”
.
Southe
ast
Asian Journal
of
Tr
opic
al
M
edi
c
ine an
d
Public
H
eal
th
.
1
;
45
(1)
:95
,
201
4
.
[3]
Kakc
hapati
S,
Yottha
noo
S,
C
hoonpra
dup
C.
“
Modeli
ng
tube
r
cul
osis
inc
id
ence
in
Nepa
l
”
.
As
ian
Bi
omed
.
Nov
8;4(2)
,
2010
.
[4]
Golub,
J.E
.
,
Sar
ac
en
i,
V.
,
Cav
alca
nt
e,
S.C
.
,
Pa
c
hec
o,
A
.
G.,
Mo
ult
on,
L.
H.
,
Kin
g,
B.
S.
,
Efr
on
,
A.,
Moore,
R
.
D.
,
Chai
ss
on,
R.
E
.
and
Durovni,
B
.
The
impact
of
ant
ir
et
rovir
al
therap
y
and
isonia
zi
d
pr
ev
ent
iv
e
the
r
ap
y
on
tube
rcu
losis
inci
denc
e
in
HIV
-
in
fec
t
ed
pa
tients
i
n
Rio
d
e
Jane
iro
,
Bra
zi
l
.
AIDS
(
London,
Engl
an
d)
,
21
(11),
p.
144
1
,
2007
.
[5]
Murra
y
CJ,
Sal
om
on
JA
.
“
Mo
del
ing
the
impa
ct
of
global
tub
erc
ulosis
cont
ro
l
stra
te
gi
es
”
.
Pr
oce
ed
ings
of
th
e
Nati
onal Ac
ad
e
my
of
S
cienc
es
.
Nov 10
;
95
(23):
13881
-
6
,
1998
.
[6]
Cohen
T
,
Murra
y
M.
“
Modeli
ng
epi
demics
of
m
ult
idrug
-
r
esista
nt
M.
tube
r
cul
osis of
heteroge
neou
s
fit
ness
”
.
Nature
medic
in
e
.
Oct
;
1
0
(10):1117
,
200
4
.
[7]
Casti
ll
o
-
Ch
avez
C,
Song
B.
“
Dy
namica
l
m
odel
s
of
tube
rcu
losis
a
nd
the
ir
app
li
c
ations
”
.
Math
emat
ic
al
biosc
ie
n
ce
s
and
engi
n
ee
ring
.
Sep
1
;
1
(2
):361
-
404
,
2004
.
[8]
Blower
SM
,
Ch
ou
T.
“
Modeli
n
g
the
emerge
n
ce
of
th
e
'
hot
zone
s
'
:
tube
r
cul
osis
a
nd
the
amplifica
ti
on
d
y
nami
cs
o
f
drug
resista
n
ce
”
.
Nature
med
icin
e
.
2004
Oct
1
;
1
0
(10):1111.
[9]
D
y
e
C,
W
il
l
ia
m
s
BG.
“
The
popula
ti
on
d
y
namics
and
cont
rol
of
tu
ber
cul
os
is
”
.
Scie
nce
.
Ma
y
14
;
32
8
(5980):856
-
61
;
2010
.
[10]
Le
e
J
,
Kim
SB
,
Park
GL.
“
Data
Anal
y
sis
for
Solar
Energ
y
Gene
r
at
ion
in
a
Univ
e
rsit
y
Microgr
id
”
.
Int
ernati
onal
Journal
of
Elec
t
rical
&
Compute
r E
ngineering
.
Jun 1
;
8
(3)
,
pp
.
2
088
-
8708
,
2018
.
[11]
Ahm
ad
T,
Arif
S,
Chaudr
y
N,
Anjum
S.
“
Epi
demiologi
c
al
Cha
rac
t
eri
sti
cs
of
Poliom
y
e
li
t
is
durin
g
the
21st
ce
ntu
r
y
(2000
-
2013)
”
.
In
te
rnational
Jour
nal
of
Public
He
alt
h
S
ci
en
ce (
IJP
HS)
.
Sep
1
;
3
(3
):143
-
57
,
2014
.
[12]
Zi
gnol
M,
Hos
seini
MS
,
W
right
A,
W
ee
ze
nb
ee
k
CL,
Nunn
P,
W
a
tt
CJ,
W
il
l
ia
m
s
BG,
D
y
e
C.
“
Gl
obal
in
ci
den
ce
o
f
m
ult
idrug
-
resist
ant
tube
rcu
losis
”
.
The
Journal
o
f infect
ious d
iseases
.
Aug 15
;
194
(
4):479
-
85
,
2006
.
Evaluation Warning : The document was created with Spire.PDF for Python.