T~p chl
Tin hoc
va
Di'eu
khi€n hoc, T.16, S.l (2000), 52-58
A ,
,c
A If
,I,
M9T CACH TIEP C~N RA QUYET f)~NH TRONG CHAN ·f)OAN
LAM SANG
DO VAN THANH
Abstract. The main purpose of this paper is to present an approach for applying aggregation model
in possibility theory proposed in the papers
[3 - 8]
in processes of clinical diagnostics with participation
of many medicine specialists.
1.
D~T VAN DE
Qua trlnh ch[n dean lam sang ngtrci b~nh la qua trlnh thOng thiro'ng diro'c tlnrc hien bO'i t~p
th~ cac chuyen gia y hoc. Day 111.
khfiu
bitt bU9Cva anh hircng quan trong dgn chat hro'ng di'eu trio
M\lC dich cua qua trlnh nay nh~m xac dinh dung benh, rmrc d9 mitc b~nh ciia ngtro'i b~nh va dira ra
bi~n phap dieu tri ban dau.
Trong qua trlnh ch[n dean, m~i chuyen gia y hoc se dtra VaGtri~u chimg Him sang ngtroi b~nh,
dira VaGtri th irc y hoc chung da diro'c t5ng ket va dira VaGtri tlnrc kinh nghiern cua chinh mlnh d€
dua ra
y
kign ch[n dean.
Nhieu tlnh huang xay ra la ngirci b~nh bie'u hi~n lam sang khong ro net, nhirng chuyen gia y

trongIy thuyet kha nang da diro'c de xuat trong cac tai li~u
[3-8].
Bai nay chi trlnh bay han chg mot khia canh img dung ciia phtro'ng ph ap tkh hop thong qua
vi~c gi&i thi~u mf hlnh tfch hop trong ch[n dean lam sang ngtrci b~nh.
2.
CO' SO' TRI THUC CAN THlET COA cAc pHAN DoAN
KHONG CH.,lC CHAN
Gii sd' c6 m chuyen gia y hoc tham gia thu'c hi~n ch[n dean lam sang nguoi b~nh. M~i chuyen
gia thirong dira ra cac
y
kien phan dean cua rninh dirci dang t~p cac cau kie'u nhir:
1) Co the' tin rhg (cUc chh rhg) ngiroi b~nh co chirng. b~nh [hoac ngtrci b~nh can diro'c]
"ten
cac
ket lu~n" •
2)
VI
nguei b~nh co cac trieu chimg
"ten
cac
tri~u clnrng"
nen co the' tin rhg (cUc chh
rhg) ngtroi d6 co chirng benh
[hoac
ngiro'i do can diro'c]
"ten
cac
ket luan".
Sau d6 t~p the' cac chuyen gia se phan tfch tat d. cac phan doan d6 de' rut ra cac phan doan
thkh ho'p nh St; Truong hop khi bie'u hi~n lam sang ngtroi b~nh khong ro net ho~c co nhieu bie'u

gan n ir c ac c an rang, a c ac c an rang, ngucn
0
co c trng en oac ngtrm
0
nen iro'c
"ten cac
kih
lu~n".
MQT CACH TIEP CA-N RA QUYET D~NH TRONG CHAN DoAN LAM SANG
53
Trong nhirng ket luan kie'u nay, cac tir nhir: co nhieu khd nang, gan nhv: ch;{c cMn, klui cMc
h" • ~ th" hi A At khf h" h~ "', h d' d" , , h' d' ~" A
can, nen
ac
oc.:
e ien rno sir ong e ae e an ve tm ung an eua cae p an oan. v
1
v~y
din phai eho each danh gia v'e cac evm tjr: co nhieu khd nang, gan nhv: ch;{c chrtn, kha chifc
.is«
Phuong phap dircc sll- dung trong nhirng trircng hop nhir v~y thtro'ng la dung cac gia tri so (ho~c
gia tri
ngon
ngir] de' iroc hrong niern tin
vao
tinh dung ditn
cua cac ph
an
doan
e6 chira

S,!
dung dlin cua cac phan doan thich hop nhat rna t~p the' chuyen gia da xac dinh ta se nhan diroc
m9t CO' s& tri thirc can thidt [gia tri khoang] trong ly thuydt kha nang
[1].
Bien nay la ffi9t sir goi
Y
de' d'e xufit
mf
hmh tich hop cac y kien chuyen gia trong qua trinh cha:n doan lam sang.
.• '
~
.•.
,
3. S(1 DO CHAN DOAN VA MO HINH TICH HQ'F
3.1.
SO'
do
qua trinh cha?n
doan
Qua trrnh eha:n doan lam sang ngiro'i benh c6 the' diro'c mo ta thong qua 5 giai doan theo so'
dt
sau
Giai dean
1
ehinh la giai dean thu th~p cac y kien phan doan cua cac chuyen gia. Giai doan
2 se loai b3 nhfi:ng phan doan vo ly kh6 e6 the' chap nhan, nhirng ph an doan dir thira. Giai dean 3
tlnrc chat la giai doan thu th~p y kien danh gia v'e niern tin cua m~i chuyen gia v'e
SV'
dung dlin ciia
cac ph an doan cua tat ca cac chuyen gia. Giai dean 4 se tich ho'p cac y kien chuyen gia de'

dung
cua
t~p
cac phan doan
(no diro'c xem
nhu
I~ me?t CO"
50'
tri
tlnrc]. M6i
y
kien
danh gia khi do thirc chat Ia mot CO" sO-tri th trc kha nang gia tr] can thiet. Vi v~y ta co thg tfch ho'p
cac
y
kien
ph
an doan theo phirong phap da. diro'c trlnh bay trong [6- 8]' c~ thg vi~c tich
hop
nay se
du'cc tien hanh theo htrtrng tiep c~n tien de h6a va diro'c
thuc
hi~n tren cac phan bo kha nang d~c
bi~t d~c tru'ng cho cac
y
kien phan dean do [6- 8].
Duo-i day se du'o'c trlnh bay m9t each tom tift
y
tU'o'ng CO" ban cu a plnro'ng phap nay: Gicl.sU-
r

nghia Ia
phan dean
Sichitc chitn se dung it nhat
voi
rmrc de?ai hay
co the' noi N(Si)
2:
ai,
0-
day N Ia me?t de?do can thiet (hay de? do chitc chitn) tren ngon ngir diroc
sinh tu: t~p
cac
ph an dean Si. Khi do F
=
{(Si, ai)
I
N(S;) 2: ai, ai
E
[0,1]'
i
=
1,
,n}, dtro'c xem
nhir Ia m9t
Y
kien chuyen gia ve rmi'c de?can thiet ctia t~p cac phan dean
r.
Doi vci y kien chuyen gia F phiro'ng phap hinh thtrc de' xac dinh phan bo khi nang d~c trtrng
cho
y

Lay
y
hen, Loei trv: stf khcic bi~t.
voi
rnoi w
E
{1
([1,7-8]).
3.3. M6 hinh tich h<!p m&-r<)ng
• Ma hmh nay thirc chat Ia me?t su' khai quat hoa md hlnh vira diro'c trinh bay tren. Neu nhu
mo hmh tren nh~m gicl.iquyet van de rich hop cac
y
kien chuyen gia,
o'
do ta chira quan tam mdt
each thoa dang den str khac bi~t ve trinh de?tri thirc, kinh nghiern cua cac chuyen gia tham gia vao
qua trinh chin doan, thl
ma
hlnh me rfmg se khitc phuc nhuoc di~m nay
[9-10].
Cu thg ta se tfch
hC?1>cac y kien chuyen gia khi mi)i chuyen gia deu diro'c gltn
vci
me?t trong S() d~ do tam quan trong,
hay do gia tri kinh nghiem, tri thirc cua chuyen gia do.
Co hai each tiep c~n de' gicl.iquyet van de nay
[9-10].
Trong
[9]
chung toi da. trmh bay each

a
=
<I>(a1'
a2, , am)
MQT CACH TIEP CA-N RA QUYET f)~NH TRONG CHAN f)oAN LAM SANG
55
trong do
m
'" ai
~(Pl,p2'
"',Pm)
= ~
2:
a
Pi
i=1
.>1
J
J_
6-day m la s5 chuyen gia eha:n dean,
PiE
[0,1] vo'i moi
i
=
1, ,m.
Tinh chat cua toan tu' nay dii diro'c chi ra trong [10], ev the' la no thoa 7 di'eu ki~n doi hoi eho
cac qua trlnh tieh hop
"Chi ph1f thuqc
stf ki4n"
cda cac phan bo xac xuat,

8
4
:
Chlte ehltn rhg dira tr~ bi suy dinh dufrng.
8
5
:
Gan nhir ehite ehltn d.ng
CO"
quan dai tie'u ti~n cua dira tr~ tot,
Th~t ra t~p cac ph an dean "nay la m9t
Sl!:
bien the' tIT m9t vi du cua Dubois va Prade
[2],
dii
dircc nghien ciru phat trie'n trong [7],
D~t:
a
=
"dira tr~ bi suy dinh dufrng":
b
=
"du-a tr~ bi da
vang,
bung ong. bieng an"; c
=
"dira tr~ bi gan yeu";
d
=
"co"quan dai tie'u ti~n cua dira tr~ t5t",

d);
W2
=
(a,b,c,-,d);
W3
=
(a,b,-,c,d);
W4
=
(a,b,-,c,-,d);
W5
=
(a,-,b,c,d);
W6
=
(a,-,b,c,-,d);
W7
=
(a,-,b,-,c,d);
Ws
=
(a,-,b,-,c,-,d);
Wg
=
(-,a,b,c,d);
WID
=
(-,a,b,c,-,d);
Wu
=

kien cda ho ve rmrc de? ean thiet doi v&i
tinh trang dung ciia cac phan doan tren diroc me ta. trong bang 1.
Bdng
1.
Y
kien cua cac chuyen gia
Chuyen gia
Phan
doan
(8d
(8
2
)
(8
3
)
(8
4
)
(8
s
)
-,a
vb
-,b
V
c
-'a
V
-,d

it nhilt dei vai mt)i y kien
d. nh
an
tren,
ta se nhan diro'c cac ph an be kH nang' d~e
d.
it nhat ttrcrng ling (being
2).
Bdng
2. Being cac phan b9 kha nang d~e trtrng eho cac y kien chuyen gia
WI
W2 W3 W4 Ws
W6 W7
Wg Wg
WlO
Wll
WI2 WI3
W14
W15 WI6
FI
0,20 0,80 0,50 0,40 0,20 0,30 0,30
0,30 0,50
0,50 0,40 0,40
0,40 0,40
0,40 0,40
F2
0,30 0,70 0,50 0,40 0,30 0,35 0,35
0,35 0,50 0,50
0,40 0,40
0,40 0,40 0,40 0,40

a. Tich ho'p phan b~e theo cac 16"pco cung thu tv tv nhien
(lien ket
veri
mqt loos todn. ttt)
Gici suor~ng ta chon toan tu
ton tronq 11kien
so
il,ong
!P
rm
la toan tu' lien ket v6i. qua trinh
tich hop phan b~e nay.
Ky hi~u
1ragg
la y kien tich hop diro'c sinh ra bch qua trinh nay, khi d6
1r
a
gg
=
!P
rm
(!p
rm
(1rF, (~), 1rF, (w), 1rF. (w)),
!P
rm
(1rF. (w), 1rF. (w), 1rF6(w)
)!p
rm
(1rFT (w)))

0,40
0,40 0,40 0,40
1r2
0,40 0,55 0,70
0,55 0,40
0,40 0,40 0,40 0,70 0,55 0,30 0,30 0,60 0,55
0,60 0,55
1rF
T
0,35
0,50 0,75 0,50
0,40 0,40 0,40 0,40
0,60
0,50 0,35 0,35 0,60 0,50 0,60 0,40
1ragg
0,40
0,80 0,75 0,55 0,40
0,40 0,40 0,40 0,70 0,50 0,40 0,40 0,60 0,55 0,60 0,40
Vi v~y trong truong hop nay y kien tich hop tir 7 y kien tren ve cac phan doan d~ cho la:
(S1,
0,6),
(S2,
0,6),
(S3,
0,6),
(S4,
0,6)' (Ss, 0,6).
b. Tich h<!p phan b~c theo cac 16"pco cling d(J mau thu.5.n
(lien ktt
vo-i

ky,
y kien nao
e6 d9 mau thuh nho hon se diro'c coi
111.
quan trong hon va dircc
U'U
tien thu nh~n
truxrc.
Bay gi<r
gia suo
1r:
gg
la ky hi~u cua ph an be ket qua nhan diro'c tir qua trinh tich hop phan b~e lien ket vai
toan tu
!Pro
&
tren, khi d6
1r:
gg
diro'c xac dinh boi
1r:gg(w)
=
!pro(
e.,
(1rF, (w), 1rF. (w), 1rF. (w)),
!Pro
(1rFT(w))'
!Pro
(1rF, (w), 1rF. (w), 1rF6(w)))
v6i. moi lap the gi6i. c6 th~

0,40
0,40 0,60 0,55 0,60 0,55
2
7r;gg
0,20 0,80 0,80 0,40 0,20 0,30 0,30 0,30 0,80 0,60 0,40 0,40
0,70
0,40 0,70 0,40
Tir bing nay ta
nhan
dtro'c
y
kien
tich hop
F:gg:
(8
1
,
0,8)'
(8
2
,
0,2),
(8
3
,
0,8),
(8
4
,
0,2),

tu-
<1>1, <1>2.
Ta c6
7r~gg(W)
=
<1>ro(7rdW),7rFr(W),7r2(W))'
va ta c6 bang sau:
Wl
W2 W3 W4 W5 W6 W7 Ws
Wg
WID
wll W12 W13
W14 W15 W16
7r~gg
0,30 0,80 0,70 0,40 0,50
0,40
0,40 0,40
0,70 0,50 0,40 0,40 0,60 0,40 0,60 0,40
Vi v~y ta
nhan
diro'c
y
kien tich
hop
F~gg:
(8
1
,
0,5)'
(8

c6 th~ suy di~n dtro'c tir
cac phan dean
n6i
tren.
Theo
y
kien cua m6i chuyen gia ve mire di? din thiet ve tinh dung dh cii a t~p
cac
phan doan
[diro'c mo tA
?y
cac
bang
1,
2) ta nh~n diro'c rmrc di? can thiet it nhat d~
"aua tre b; gan ylu"
ttro'ng
iing
vo'i
7
y
kien
chuyen
gia d6
u
0,6; 0,6; 0,55; 0,4; 0,3; 0,4; 0,4.
Neu
y
kien chung dtroc sinh b?Yiqua trinh tich ho'p
phan

Neu
y
kien chung diro'c sinh b?Yiqua trlnh tich ho'p ph an b~c sd- dung h~n hop hai loai toan td-
tich hop la toan td- ton trong
y
kien
si5
dong va toan td- ton trong tlur t~, thi theo
y
kien nay mire
di? can thiet it nhat
M
phan dean
"aua tre bi gan ylu"
khOng nho hon 0,4.
5.
KET
LU~N
Van de xay dung cac h~ chuyen gia va h~ tro' giup quyet dinh trong chin doan b~nh vm cac
tri tlnrc day dil, chitc chitn cila chuyen gia dii dtro'c quan tam nghien
CU:u
tir cuoi th~p k170 va dii c6
san ph am thirong mai, nhimg voi cac tri thirc dircc biet khong day dd, khOng chitc chitn vh chira
co san phitm thiro'ng mai chfnh thirc m~c du n6 du'oc quan tam nghien ctru rat manh trong vai narn
~d~ .
Thirc te trong cac qua trinh chin dean va dieu tri, thircng hay xay ra trirong hop la nhieu
b~nh nh an c6 tri~u chimg virot qua kha nang chitn doan chinh xac cua hac
S1
dieu tri va cling thirong
58

TAl
L~U
THAM KHAO
[1] D. Dubois,
J.
Lang, H. Prade, Possibilistic Logic,
Handbook of logic in Artificial Intelligence
and Logic Programming, Volumme 9, Nonmonotonic Reasoning and Uncertain Reasoning, Eds.'
Dov. M. Gabbay, C.
J.
Hogger,
J.
A. Robinson, D. Nute, Clarendon Press, Oxford, 1994, 438-
510.
[2] D. Dubois and H. Prade, Epistemic entrenment and possibilistic logic,
Artificial Intelligence
50
(1991) 223-239.
[3] D. V. Thanh, A relationship between the possibility logic and the probability logic,
Computer
and Artificial Intelligence
17 (1) (1998) 51-68.
[4] D. V. Thanh, Stability of the principle of minimal Specificity and maximal Buoancy,
Tq.p cM
Tin hoc va Dieu khie'n hoc
12,
No.4 (1996) 1-17.
[5] D. V. Thanh, Application of Stability of the principle of minimal Specificity and maximal
Buoancy, accepted for oral presentation in
The Joint Pacific Asian Conference on Expert Sys-

12 - 7 -1 998
Van phong Ban cM doo ChuO'ng trinh quoc gia
ve Cong ngh~ thOng tin