Developing Intelligent Health Diagnosis System for Korean Traditional Medicine: Database Construction and Neural Inferences

In this paper, we propose a self-health diagnosis system based on Korea Traditional Medicine that has thousand years of history and popular in Korean general public. The system requires constructing a reliable diseases-symptoms database and classification learning method. Database construction is based on various reports submitted to the government about “Diseases burdensome to Korean Patients” in 2005 and medical contents “Engel Pharm” with 60 diseases. An enhanced ART2 algorithm that has dynamic control of boundary variables to control the number of clusters inside is used to extract top five most probable diseases generated by simple user input. The constructed database and test diagnosis results are verified by Korean Traditional medical doctors and the experiment also shows that the system is easily accessible and reliable in accuracy.