FUZZY EXPERT SYSTEM FOR DIABETES USING REINFORCED FUZZY ASSESSMENT MECHANISMS M.KALPANA

Transcription

1 FUZZY EXPERT SYSTEM FOR DIABETES USING REINFORCED FUZZY ASSESSMENT MECHANISMS Thesis Submitted to the BHARATHIAR UNIVERSITY in partial fulfillment of the requirements for the award of the Degree of DOCTOR OF PHILOSOPHY IN COMPUTER SCIENCE By M.KALPANA Under the Guidance of Dr. A.V. SENTHIL KUMAR M.C.A, PGDCA, M.Phil, Ph.D., Professor and Director POST GRADUATE AND RESEARCH DEPARTMENT OF COMPUTER APPLICATIONS HINDUSTHAN COLLEGE OF ARTS AND SCIENCE Hindusthan Gardens, Behind Nava India, Coimbatore Tamil Nadu, India January 2014

2 DECLARATION I M.KALPANA hereby declare that the thesis, entitled Fuzzy Expert System for Diabetes using Reinforced Fuzzy Assessment Mechanisms submitted to the Bharathiar University, in partial fulfillment of the requirements for the award of the Degree of Doctor of Philosophy in Computer Science is a record of original and independent research work done by me during January January 2014 under the Supervision and Guidance of Dr. A.V. Senthil kumar, Professor and Director, Department of Post Graduate and Research Department of Computer Applications, Hindusthan College of Arts and Science, Coimbatore and it has not formed the basis for the award of any Degree / Diploma / Associateship / Fellowship or other similar title of any candidate of any university. Countersigned (Research Supervisor) Signature of the Candidate i

3 CERTIFICATE This is to certify that the thesis, entitled Fuzzy Expert System for Diabetes using Reinforced Fuzzy Assessment Mechanisms submitted to the Bharathiar University, in partial fulfillment of the requirements for the award of the Degree of Doctor of Philosophy in Computer Science is a record of original research work done by M.KALPANA during the period January January 2014 of her research in the Department of Post Graduate and Research Department of Computer Applications at Hindustan College of Arts and Science, Coimbatore under my supervision and guidance and the thesis has not formed the basis for the award of any Degree / Diploma / Associateship / Fellowship or other similar title of any candidate of any university. Counter signed Research Supervisor Principal (Hindustan College of Arts and Science, Coimbatore) ii

4 CERTIFICATE OF GENUINESS OF THE PUBLICATION This is to certify that Ph.D. candidate M.KALPANA working under my supervision has published a research article in the standard referred/sci journal named European Journal of Scientific Research with Vol. 97 No. 1 Page Nos and year of publication 2013 published by European Journals Inc. The contents of the publication incorporates part of the results presented in her thesis. Counter signed Research Supervisor Principal (Hindustan College of Arts and Science, Coimbatore) iii

5 ACKNOWLEDGEMENT Foremost, I would like to express my sincere gratitude to my supervisor Dr. A.V.Senthil Kumar, Professor and Director, Post Graduate and Research Department of Computer Application for the continuous support of my Ph.D study and research, for his patience, motivation, enthusiasm, and immense knowledge. His guidance helped me in all the time of research and writing of the thesis. I would like to express my deepest gratitude to Dr.N.Balusamy, Principal, Hindusthan College of Arts and Science for his excellent guidance, caring, patience, and providing me with an excellent atmosphere for doing research. I would like to offer my special thanks to Dr.P.Pandiyarajan, Dean, Anbil Dharmalingam Agricultural College and Research Institute, Tiruchirappalli, Dr. N.R.Padmanaban, Professor and Head, Department of Social Sciences, Anbil Dharmalingam Agricultural College and Research Institute and the faculty members of Social Sciences department for their encouragement and motivation. I am most grateful to Dr.S.Selvam, Professor (Agricultural Economics) for his valuable comments for writing the thesis. My sincere thanks to Dr.V.Manonmani, Professor (Seed Science and Technology) for encouragement and insightful comments. I would like to thank Dr. S.Chitra Assistant Professor (Plant Breeding and Genetics) who as a good friend supported me to implement my work in the field of Agriculture. I would like to thank my friend K.Rajeswari for her motivation to complete my research. Last but not the least, I would like to thank my family: my husband K.Sivakumar, my daughter S.K.Thejashwini, my parents A.Muthusamy, M.Rajeswari and my brothers A.M. Saravanan and A.M Venkatesh for supporting spiritually and encouraging me at each stage of my research work. (M.KALPANA) iv

6 ABSTRACT Decision making has its own role in all the areas. Especially in the field of medicine, diagnosis is very much essential to assign the patients at any of group (not at risk, less risk, at risk). Medical practitioners exhibit variation in decision making to deal with uncertainties and ambiguity in knowledge and information. Fuzzy logic presents powerful reasoning methods that can handle uncertainties and vagueness. Fuzzy Expert System is an artificial intelligence system which helps to solve the decision making problem with the existence of uncertainty. The Fuzzy Expert Systems (FES) define imprecise knowledge and offers linguistic concept with excellent approximation. FES plays an important role in medicine for symptomatic diagnostic remedies. FES provides a natural way to include human expertise in the form of If-Then decision rules, based on and very close to the linguistic description of the human expert. In the field of medicine there exist high risk disease such as heart disease, diabetes, high blood pressure and cancer. Diabetes is the high risk disease that arises day by day in India. So Fuzzy Expert System was proposed in this study to develop an early detection and assessment tool for diabetic people. The proposed FES is constructed using the algorithm Reinforced Fuzzy Assessment Mechanisms. In general, FES is constructed with Fuzzy Inference Mechanism which uses fuzzy sets, membership function, uncertainties are managed with Certainty Factors and rules are constructed using fuzzy sets. Reinforced Fuzzy Assessment Mechanisms holds the algorithms such as Fuzzy Assessment Methodology; Enhanced Fuzzy Assessment Methodology; T Fuzzy Assessment Methodology and S Fuzzy Assessment Methodology. Fuzzy Assessment Methodology using Fact Values to find the uncertainty using evidence and hypothesis. The Fact Values (FV) derived helps to handle uncertainty in rules. Correlation Fuzzy Logic is used to find the relationship between fuzzy numbers and membership function. The parameters for the membership function are fixed by using the methodology called MMMSDV (Minimum, Maximum, Mean, and Standard Deviation Values). Enhanced Fuzzy Assessment Methodology helps to find the overlapping between membership function using K ratio. Membership function and overlapping between the membership function is very important part in Enhanced Fuzzy Assessment Methodology. To find the overlapping between the membership function, K ratio was derived with certain criteria. K ratio finds whether the membership function overlap between each other or not. x

7 T Fuzzy Assessment Methodology is used to measure similarity between fuzzy sets, fuzzy numbers and fuzzy rules. If there are similar fuzzy set and rules, the computation time of FES takes long time. So similarity measure is very essential in FES. S Fuzzy Assessment Methodology helps to measure uncertainty in data. S Value helps to find the uncertainty in data by using the fuzzy rules. With T Value and F Value, we compute S Value using fuzzy rule to manage uncertainty. The proposed algorithm was tested with Pima Indian Diabetes Dataset and implemented with MATLAB Fuzzy Logic Toolbox. The final result from the MATLAB Fuzzy Logic Toolbox was given to the medical practitioner for interpretation. Performance Assessment of the proposed algorithm can be calculated based on Accuracy level. The Accuracy of the proposed FES is very efficient compared to the existing method. The results obtained from FES are crisp values. These values cannot be directly interpreted. They are converted into knowledge. To convert into knowledge we have two parts. They are Statement Study and Assessment Statement. In Statement Study the data for single person is exhibited. This part gives the overall parameter about a patient. Assessment Statement gives the results of patient suffering from diabetes (low or verylow, medium and high or veryhigh). The algorithm was tested for all the Fuzzy Association: Age group. The final enhanced algorithm was tested in another domain ie., in the field of agriculture to diagnosis the yield of rice. The algorithm works more effectively in another domain. xi