基于实体识别的医嘱复合词标准化方法的应用研究

Application research of standardization method of medical advice compound words based on entity recognition

目的:提高医嘱复合词数据清洗和数据标准化的效率。方法:收集2015年―2021年在解放军总医院急救患者的住院医嘱数据,共计22809人,27096例次。以知识库为基础,利用分级词典将医嘱复合词拆解为不同层级的实体,基于语义重组算法对实体进行笛卡尔积重组,重组结果做标准化处理,最终辅以临床医学专家对标准化处理结果人工校验。结果:针对医嘱复合词,如检查和手术类医嘱的标准化效果良好。结论:本方法提高了医嘱复合词标准化的准确率,节省了数据治理的时间。

Objective To improve the efficiency of data cleaning and data standardisation of medical advice compound words.Methods The inpatient medical advice data of emergency patients in PLA General Hospital from 2015 to 2021 were collected,a total of 22,809 patients and 27,096 cases.Based on the knowledge base,the hierarchical dictionary was used to disassemble the compound words of medical advice into entities of different levels,and the entities were reconstructed by Cartesian product based on the semantic reorganization algorithm.The results of the reorganization were standardized,and finally the standardized processing results were verified by clinical medical experts.Results The standardization effect is good for the compound words of medical advice,such as examination and surgery.Conclusion This method can improve the accuracy of the standardization of medical advice compound words and can save the time of data governance.