1. UKM Learning and Research Repository
2. PHD Thesis / Tesis PHD
3. Faculty of Science and Technology / Fakulti Sains dan Teknologi

Title:	Disease diagnosis: a fuzzy approach
Authors:	Asaad A. Mahdi (P38441)
Supervisor:	Ahmad Mahir Razali, Associate Professor Dr.
Keywords:	Disease diagnosis Diagnosis disease Fuzzy approach Fuzzy systems in medicine
Issue Date:	13-Dec-2012
Description:	Information technology could be used to reduce the mortality rate caused by diseases affecting millions of people every day while waiting to get better treatment from the specialists. Fuzzy set theory and fuzzy logic are a highly suitable basis for developing knowledge based systems in medicine for tasks such as diagnosis of diseases, and the optimal selection of medical treatments. The goal of this thesis is to develop a methodology using fuzzy set theory to assist general practitioners in diagnosing and predicting the patients' condition from certain rules based on experience. Medical practitioners other than specialists may not have enough expertise or experience to deal with certain high risk diseases. With this diagnosis the patients with high risk factors or symptoms could be short listed to see the specialists for further treatment. The fuzzy diagnosis is based on the doctors' ability to make an initial judgment gained from his study and experience. In this thesis the following steps were taken, (1) Choosing diseases and designing a unique questionnaire to be used to collect the data needed. (2) Collecting the data from a random sample taken from different general and private clinics and hospitals in Kuala Lumpur, sample size taken for the study was 200 patients. (3) Constructing three different data sets. The first set is the disease symptoms set S ={s1, s2 , s3 , s4 ,..., sk}, which contain information on symptoms of diseases. The symptoms were fever, high temperature, headache, pain behind eyes, nausea, vomiting, rash, joint pain, muscle pain, bleeding, lethargic, loss of appetite, diarrhea, cough, chest pain, chills, sore throat, itchiness, abdominal pain and runny nose. The second set is the diseases set which consists of all diseases that have similar symptoms D = { d1, d2,..., dm} such as chicken pox, hepatitis B, measles, dengue and flu, and finally the patients set, P = { p1, p2, p3..., pn} . (4) Forming the unique membership functions for each symptom, building max-min composite binary fuzzy relations equations and calculating fuzzy relations. The data were then tested for multicollinearity and reliability. Factor analysis was conducted to check the symptoms loading onto factors. A special program was written to calculate the occurrence and confirmability measures by using the expert knowledge about the diseases. (5) Running the diagnosis on a sample combination of two, three, four and five diseases and compares the results with the doctors' diagnosis. The results of diagnosis were varied from excellent, 97% for the diagnosis of the combination of measles and dengue, very good 83% for the diagnosis of the combination of hepatitis B, measles and flu, and good 79% for the diagnosis of the combination of measles, flu and chicken pox. Finally, using the measures of accuracy to check the diagnostic accuracy and compare the performance of fuzzy, K- Nearest Neighbor and Naive Bayes classifiers as methods of disease diagnosis, it was found that fuzzy classifier outperforms the other methods in all cases of diseases samples.,PhD
Pages:	226
Call Number:	QA9.64 .A464 2012
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Science and Technology / Fakulti Sains dan Teknologi

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.