Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Web-Based Expert System To Diagnose Heart And Lung Diseases Using Fuzzy Logic
And Certainty Factor
(International Journal of Knowledge-Based and Intelligent Engineering Systems)
(Website: http://www.iospress.nl/journal/international-journal-of-knowledge-based-and-intelligent-engineering-systems/ )
Oleh:
Erna Yulianti
(1104505008 )
(1104505047)
(1104505050)
(1104505060)
(1104505066 )
1.
Introduction
Research method
Body temperature.
This input variable has divided into 5 fuzzy
sets: Very Low, Low, Normal, High and
Very High. Membership function of Low,
Normal and High sets are triangular, while
membership function of Very Low and Very
High sets are trapezoid. These fuzzy sets will be
shown in Table 1. Membership functions of body
temperature field will be shown in Fig. 1.
Table 1. Classification of Body Temperature
Input Variable
Range
Body
Temperature
<360 C
350C 370C
360C 380C
37.50C
39.50C
>38.50C
Fuzzy
Sets
Very
Low
Low
Normal
High
Very
High
Pulse Rate
0 50
30 70
50 100
90 140
>120
Very Low
Low
Normal
High
Very High
36-x
VeryLow ( x ) = 1.0 , 35x36
1, x35
Low ( x )
50-x
VeryLow ( x ) = 20 , 30x50
1, x30
x-35
, 35x36
1.0
= 1, x=36
37-x
, 36x37
1.0
Low ( x )
{
{
x-36
, 36x37
1.0
Normal ( x ) = 1, x=37
38-x
, 37x38
1.0
High ( x )
b.
Input Variable
Interva
l
Fuzzy
Set
{
{
x-50
, 50x75
25
Normal ( x ) = 1, x=75
100-x
, 75x100
25
x-37.5
, 37.5x38.5
1.0
High ( x ) = 1, x=38.5
39.5-x
, 38.5x39.5
1.0
Pulse rate.
Pulse rate variable has 5 fuzzy sets: Very
Low, Low, Normal, High and Very
High. Membership function of Low,
Normal and High sets are triangular, while
membership function of Very Low and Very
High sets are trapezoid. These fuzzy sets will be
shown in Table 2. Membership functions of pulse
rate field will be shown in Fig.2.
x-30
, 30x50
20
= 1, x=50
70-x
, 50x70
20
x-90
, 90x115
25
= 1, x=115
115-x
, 115x140
25
x-120
, 120x140
VeryHigh ( x ) = 20
1, x140
c.
x-140
, 140x150
VeryHigh ( x ) = 10
1, x150
Input Variable
Systolic
Pressure
Interval
Blood
<50
40 100
90 140
120
160
>150
Fuzzy
Set
Very
Low
Low
Normal
High
Very
High
d.
Body weight.
Body weight variable is divided into two
categories based on the gender, adult female and
adult male. Each category has 5 fuzzy sets
(Very Low, Low, Normal, High and
Very High). Membership function of Low,
Normal and High sets are triangular, while
membership function of Very Low and Very
High sets are trapezoid. These fuzzy sets will be
shown in Table 4. Membership functions of
female body weight field will be shown in Fig. 4.
Table 4. Classification of adult female body weight
Input Variable
Adult Female
Body Weight
Interval
<40
35 47
45 162
60 67
>65
Membership
pressure:
functions
of
systolic
Fuzzy Set
Very Low
Low
Normal
High
Very High
blood
1, x40
VeryLow ( x ) = 50-x
, 40x50
10
Low ( x )
x-40
, 40x70
30
= 1, x=70
100-x
, 70x100
30
{
{
x-90
, 70x115
25
Normal ( x ) = 1, x=115
115-x
, 115x140
25
High ( x )
x-120
, 120x140
20
= 1, x=140
160-x
, 140x160
20
1, x40
VeryLow ( x ) = 35-x
, 35x50
5
Low ( x )
{
{
x-35
, 35x40
7
= 1, x=40
47-x
, 40x47
7
x-45
, 45x47
15
Normal ( x ) = 1, x=47
62-x
, 47x62
15
High ( x )
x-65
, 165x67
VeryHigh ( x ) = 2
1, x67
High ( x )
Fuzzy Set
<40
35 47
45 162
60 67
>65
Very Low
Low
Normal
High
Very High
x-60
, 60x63
22
= 1, x=63
85-x
, 63x85
22
x-80
, 80x85
VeryHigh ( x ) = 5
1, x85
Input
Variable
Adult Male
Body Weight
{
{
x-53
, 53x60
3
Normal ( x ) = 1, x=60
63-x
, 60x63
3
x-60
, 60x62
5
= 1, x=62
67-x
, 62x67
5
e.
Blood sugar.
Blood sugar variable divided into 5 different
fuzzy sets: Very Low, Low, Normal,
High and Very High. The membership
function of Very Low and Very High sets are
trapezoidal while Low, Normal and High
sets are triangular. Table 6 below depict the
fuzzy set classification for this variable. Figure 6
is the graphical representation of the membership
function curve.
Table 6. Classification of blood sugar
Input Variable
Interval
Fuzzy Set
Blood Sugar
< 70
Very Low
60 130
Low
120 180
Normal
170 240
High
> 230
Very High
1, x42
VeryLow ( x ) = 45-x
, 42x45
3
Low ( x )
x-42
, 42x45
15
= 1, x=45
60-x
, 45x60
15
1, x60
VeryLow ( x ) = 70-x
, 60x70
10
Low ( x )
{
{
{
x-60
, 60x70
60
= 1, x=70
130-x
, 70x130
60
x-120
, 120x130
50
Normal ( x ) = 1, x=130
180-x
, 130x180
50
High ( x )
x-170
, 170x180
60
= 1, x=180
240-x
, 180x240
60
x-230
, 230x240
VeryHigh ( x ) = 10
1, x240
3.
Experiment
4.
Conclusion
doctor diagnosis i
Average results of the doctor's diagnosis =
N
i=1
diagnosis
Average results of diagnosis system =
i=1
References
[1] Adeli, Ali & Mehdi, Neshat.A Fuzzy
Expert System for Heart Disease
system iDiagnosis.IAENG Proceeding of the
International
Multi
Conference
of
Engineers and Computer Science Vol. 1.
March 2010.