Expression of Laboratory Examination Results in Medical Literature

Medical literature contains expressions of laboratory examination results, which are invaluable knowledge sources for building a disease knowledge base that covers even rare diseases. In this study, we analyzed such expressions of disease descriptions in open databases with manually built dictionaries and obtained the following results. First, we identified two major types of expressions for laboratory examination results, that with and without their test names in the expressions. Second, the study identified evaluative expressions that frequently appear in the description of the results. Third, presence of test names and evaluative expressions could classify the expressions into four major classes that demand independent strategies to interpret. The study illustrated that this is beyond the scope of the existing corpora in this domain mostly designed for medical records. Although the analysis is based on rudimentary statistics, it clarified the factors necessary for future corpus designs to promote further research.