Abstract
Clinical pathways are a fixed list of viewpoints to describe patients' condition. An outcome of a clinical pathway consists of several assessments of observation items and inspection items, all of which represented as numerical data. On the other hand, free description medical records in the SOAP format are valuable data that indicate health care worker's assessments. Standardization of textual description is a key issue when integrating these types of medical data. This study extracts characteristic word 2-grams as source of sentence templates. We applied SVM (Support Vector Machine) using each outcomes as positive data. We constructed a search engine for 123736 pain variance outcomes. We conducted empirical experiments to confirm the usefulness of the proposed method.