Abstract: Structured electronic medical records can understand that it finds suitable keyword in free text, current method of extracting keyword in free text has entity recognition and the Chinese word segmentation. Structured electronic medical records find suitable keyword in a variety of semantic association, and entity recognition and the Chinese word segmentation can’t achieve its goals. Therefore, on the basis of the existing structural elements database, it establishes an inverted index in order to find a group of the most suitable simple elements combination. It gets optimum combination by using multi-objective dynamic. It reuses intersection operation method to find complex elements so as to find out a group of optimal sequences that structure the electronic medical records. Through the system implementation this method can find simple elements, and has certain semantic association between simple elements.

Key words: electronic medical record, structural element, entity recognition, inverted index, multi-objective dynamic programming

摘要： 电子病历结构化可以理解为在自由文本中找到合适的关键字，目前在自由文本中提取关键字的方法有实体识别和中文分词，但由于电子病历结构化是要在多种语义关联下找合适的关键字，而实体识别和中文分词不能达到其目标。因此根据已有的结构元素数据库建立倒排索引找到一组最合适的简单元素组合，利用多目标动态规划进行最优化组合，利用求交集方法找到复杂元素从而找出最优的一组序列即为该电子病历的结构化。通过系统实现得到利用该方法不仅可以找到结构元素，而且找到的结构元素之间有一定的语义关联。

关键词: 电子病历, 结构元素, 实体识别, 倒排索引, 多目标动态规划