Information Technology Reference
In-Depth Information
Table 25.8
Linguistic descriptions for “normal” diseases
Bronchitis
/
Pneumonia
Bronchitis Pneumonia
wheezing
sometimes occasionally
breathlessness
very seldom
often
anosmia
seldom
seldom
f ever
sometimes
often
sinusitis
sometimes
sometimes
chronic otitis
seldom
seldom
Table 25.9
Corresponding numerical descriptions for “normal” diseases Bronchitis and Pneu-
monia (Table 25.8)
d
1
d
2
s
1
0.4 0.6
s
2
0.2 0.75
s
3
0.25 0.25
s
4
0.4 0.75
s
5
0.4 0.4
s
6
0.25 0.25
that, to define such case, we need a crisp information. It is a clear situation, but is
not always a case.
The difficulty to suspect the RD is because of its symptomatic similarity to the
“normal cases”. Therefore, another alarm can be represented by possible diagnoses.
If a physician has enough experience with a computer program, s/he usually has
been working, and s/he mentions that the values for possible diagnoses are much
below or much above the usual representation, it can be also the sign of alarm “a
rare disease” for a physician.
We need to define appropriate operations and thresholds (that allow to detect
a deviation from the “normal” case). If we take max-min composition of fuzzy
relations (25.6), it restricts the maximum membership degree of possible diseases
for a patient by values in Table 25.9. For example, if a patient at hand has
S
p
=
{
(
s
1
,
0
.
4
)
,
(
s
2
,
0
.
2
)
,
(
s
3
,
0
.
25
)
,
(
s
4
,
0
.
4
)
,
(
s
5
,
0
.
4
)
,
(
s
6
,
0
.
25
)
}
(25.9)
the possible diagnosis
d
1
has the membership degree 0
.
4 (see (25.11)). If a patient
at hand has
S
p
=
{
(
s
1
,
1
.
0
)
,
(
s
2
,
1
.
0
)
,
(
s
3
,
0
.
2
)
,
(
s
4
,
0
.
25
)
,
(
s
5
,
0
.
2
)
,
(
s
6
,
0
.
2
)
}
(25.10)
the possible diagnosis
d
1
has the membership degree 0
.
4 as well.