Medical inference system on arthritis using fuzzy relational theory / Lim Chee Kau

Fuzzy relational theory is a branch in the study of fuzzy sets theory. According to this theory, mapping among two sets can be carried out in four ways, which arc represented by circle product, sub-triangular product, super-triangular product and square product. Each of these products can be abstracted into inference structures. which is useful in designing the inference engine of an expert system. Based on the revised version [DeBaets and Kerre 1993. Hallam and Yew l998a] of fuzzy relational theory. a series of 18 inference structures, namely sub-K inference structures (denoted as Kt to Kl8) are constructed. Based on these inference structures, a medical expert system is developed. This Arthritic disease diagnosis system is a two-level diagnosis system. The first level of diagnosis diagnose a patient according to the distribution of abnormalities in hands and wrists, whereas the second level diagnose a patient according to the signs and symptoms of a patient. It is designed in such a way so that level 1 diagnosis may short list possible diseases for level 2 and reduces the system work loads. The system has been tested with a number of patients and found that most inference structures shows good results in both level. Among all, inference structures K2 and K 16 show the best result in level I and level 2 diagnosis respectively