Traditional Chinese Medicine (TCM) is a discipline typically characterized as complicated information science [1], which represents the distinctive thought (e.g. theories drawn from clinic should be applied in clinic), compared with the modern medicine. Symptoms are basic clinical concepts in TCM electronic health record, which are plain text language. Nevertheless, TCM concepts are inherently characterized as semantic complexity and diversity [2], clinical terminologies related to TCM concepts have no unified standard, clinical information acquisition and expression are limited to disease-defined concepts, or the complex relationship between concepts. It is still an outstanding question that TCM symptoms are formally represented by computerized natural language processing, their semantic meaning and relationships between them are still unsolved task, so huge amounts of valuable symptom information in real world has not been fully identified and used, deep learning from TCM symptoms is difficult to be accomplished within short period.
展开▼