Mining Association Rules on Qing Court Medical Records: Semantic Abstraction and Standardization

To explore the association relations among disease, pathogenesis, physician, symptoms and drug, we adapt a variational Apriori algorithm for discovering association rules on a dataset of the Qing Court Medical Records. There are five types of semantic associations we intend to discover, including Disease-Pathogenesis-Drug set(DPaD), Disease-Symptoms-Drug set (DSyD), Disease-Drug set (DD), Disease-Physician-Drug set (DPhD) and Disease-Drug Category Set (DDC). To solve the synonymity problem and the data sparseness problem, we give a mapping strategy which maps pathogenesis to standardized forms and maps drugs to drug categories. With the mapping strategy the number of frequent drug sets rises from 287 to 1184. The experimental results indicate that our method with the mapping strategy is an effective way to acquire valuable semantic association rules.