Using Fuzzy Logic in Diagnosis of Tropical Malaria
1. .
A FUZZY KNOWLEDGE BASED SYSTEM FOR
CLINICAL DIAGNOSIS OF TROPICAL FEVER
Ismael SEKIZIYIVU
Master’s Thesis Defense
Thesis Committee:
Asst Professor Murat İSKEFİYELİ (Advisor)
Asst Professor .Ali GÜLBAĞ (Co.-Advisor)
Asst Professor . Mehmet Recep BOZKURT(External Faculty)
Department of Computer and Information Engineering
Sakarya University
November 14, 2014
3. Sub-Saharan Africa
Geographically, the area of the continent of
Africa that lies south of the Sahara Desert.
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
4. Tropical fever
Malaria and typhoid fever are
the major tropical fever infection
Malaria is caused by mosquitoes
Typhoid fever is caused
by Salmonella typhi.
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
5. Countries and areas at risk of
malaria transmission, 2011
In the tropics ,
malaria
approximately
causes 3,000
deaths each
day
Introduction - Problem Statement -TROPFEV system - Development Process - Conclusion
6. Geographical distribution of
typhoid, 2011
In sub-Saharan
Africa it is
estimated to
cause
725 cases
and 7 deaths
per 100,000
person-year
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
7. Knowledge Based Systems
A KBS uses knowledge embedded in a knowledge base
to solve complex problems.
A knowledge-based system has at least one and usually
two types of sub-systems,
1. A knowledge base that represents facts about the
world.
2. The inference engine that represents logical
assertions and conditions about the world.
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
User Interface
Explanation
Part
Inference
Engine
Knowledge
Acquisition
Knowledge
Base
Comput
er and
User
8. Knowledge based reasoning
techniques
There are a number of knowledge based reasoning
methods ;
1. semantic network
2. Artificial neural networks
3. case based reasoning
4. rule based reasoning.
5. fuzzy logic which was used in this project
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
9. Fuzzy Logic
“Fuzzy logic” refers to a logic of approximation.
Boolean logic assumes that every fact is either
entirely true or false.
Fuzzy logic allows for varying degrees of truth.
It can be used to represent vague and imprecise
ideas, such as “mild”, “high” or “severe”.
It uses natural words in the place of numerical
values
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
10. Fuzzy Logic
It uses fuzzy set theory that allows all values of a
function in the defined interval [0, 1]
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
11. Related work
The first medical system to use fuzzy logic was
CADIAG-2 . It was developed by Prof. K.-P.
Adlassnig and his colleagues at the University of
Vienna Medical School from the early 80's .
Today it is a central subject of research at the
Institute for Medical Expert and Knowledge-
Based Systems at the Medical University of
Vienna
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
12. Problem Statement
Laboratories ,
modern hospitals and medical
experts are scarce in rural areas
Where 70% of the population
live
Clinical diagnosis is mostly used due to scarcity
of laboratories.
The two diseases have various diagnosis features
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
13. Problem Statement
Some Symptoms of malaria and typhoid are similar
and hence a task in medical diagnosis .
Using signs and Symptoms
We develop a
system that will help in easier
Decision making and
classification during diagnosis
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
14. The TROPFEV System
The system can
be used by both
Doctors, medical
Workers and has
a easy user
interface
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
15. TROPFEV System Development Processes.
1. Fever domain knowledge source identification
2. Fever knowledge acquisition
3. Fever knowledge representation
4. Designing a fuzzy inference system
5. Implementation of TROPFEV fuzzy inference
system
6. Verification and testing
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
16. TROPFEV System Development Process.
Knowledge source
Fever knowledge representation
Fuzzy inference system design
FIS implementation & Interface design
TROPFEV Verification and testing
Fever knowledge Acquisition
Is it
Satisfactory
?
Roll out
Fever knowledge Acquisition
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
17. 1.knowledge source identification
Document using the Uganda
clinical guideline 2012
Consulting expert in tropical
medicine here
Dr. Akusa Yuma Darlington was
consulted
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
18. 2.Fever knowledge Acquisition
Tropical Fever
Typhoid Malaria
Uncomplicated Uncomplicated
malaria
Complicated Complicated
malaria
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
Malaria and
typhoid fever
appear as
complicated
or
Uncomplicate
d
19. Fever knowledge Acquisition
S/N Code Attribute (Symptom) Category
AGE Age
Group category
PRE Pregnancy
FEV Fever
Symptoms
APP Loss Of Appetite.
CON Convulsions
VOM Vomiting
MEN Altered Mental State
PRO Prostration
ANE Anemia
DEH Dehydration
BRE Difficulty in Breathing
THR Threatening Abortion
CHI Chils
PAI Pain
SPL Splenomegaly
HEA Headache
BRA Relative Bradycardia
ABD Abdorminal Pain
MAL Malaise
COP Constipation
GUP Gut Perforation
21
Diagnosing
features of
Fever where
acquired
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
21. 4.Designing the TROPFEV
Fuzzy inference system
Defining system input and output variables.
Linguistic variables and membership functions
Defining fuzzy rules of the system
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
22. Linguistic variables and membership
functions
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
23. Membership functions
Graph showing some of the input membership
functions
Input output
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
24. Defining fuzzy rules of the
system
These rules in the rule base were created
considering all the possible circumstances
and the conditions that were mentioned in
the Uganda clinical guidelines 2012 for
both malaria and typhoid fever in their
complicated and uncomplicated form
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
25. Implementation of TROPFEV
fuzzy inference system
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
The system
was
implemented
in Matlab
2012a
26. Inserting rules in the Rule Editor
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
27. Designinterface in Matlab GUI
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
28. Verification and testing
Fever cases for 20 patients collected from Arua
regional referral hospital in northern Uganda
with the help of a medical expert has been used
to test to the system. And compare the diagnosis
of the real expert with that of the TROFEV
system.
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
31. Testing the TROPFEV system
using user Interface
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
32. Scatter plot for doctor’s
Diagnosis against TROPFEV
showing a positive correlation
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
33. Conclusion
Knowledge for knowledge based systems is limited from
only experts but can also be obtained from documents.
Fuzzy logic is good in systems where data is not available.
Rules are easy to be edited or added whenever new
knowledge is attained though it is tiresome when they
come so many.
The use of fuzzy logic in medical diagnosis can be more
emphasized for its accuracy.
The need for tropical fever decision support systems in
tropical medicine is vital.
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
34. Future work
The system can also be integrated by adding
laboratory tests for malaria and typhoid fever as
well as the therapy part.
Constructing such systems as web based medical
expert systems can save many lives of people
including tourists and remote based patients.
Introduction -Problem Statement -TROPFEV system -Development Process- Conclusion
35. Thank you
Special thanks to
Examination committee
1. Asst Professor Murat İSKEFİYELİ (Advisor)
2. Asst Professor .Ali GÜLBAĞ (Co.-Advisor)
3. Asst Professor . Mehmet Recep BOZKURT
And the audience at large
QUESTIONS ???