Abstract
This paper presents an algorithm for the diagnosis of dental diseases based on fuzzy reasoning.
Two relationships between symptoms and diseases are obtained from clinical data by using factor analysis: one is the relation between the symptoms observed and the diseases, and the other is that between the symptoms unobserved and the diseases. These relationships are fuzzily represented by linguistic truth values after the modification by dental experts. The degree of the symptoms observed in a clinic is also represented in terms of linguistic truth values.
The diagnostic results are expressed linguistically according to the degree of inference. If the number of the observed symptoms is not enough to infer, the instructions to inspect more symptoms are given. Further, if the observation of a symptom is doubtful from a fuzzy logical point of view, the algorithm gives a warning to a doctor and/or suggests him to reinspect such and such a symptom. The diagnosis is carried out by a computer for actual data. The rate of accuracy is about 85%.
