基于CRF模型的临床医嘱信息实体识别方法应用研究  

APPLICATION OF ENTITY RECOGNITION METHOD OF CLINICAL MEDICAL ORDERS INFORMATION BASED ON CRF MODEL

作　　者：宓林晖 袁骏毅[1] Mi Linhui;Yuan Junyi(Shanghai Chest Hospital,Shanghai 200030,China)

机构地区：[1]上海市胸科医院,上海200030

出　　处：《计算机应用与软件》2020年第3期209-212,共4页Computer Applications and Software

摘　　要：将命名实体识别技术运用于临床业务系统中,实现对临床医嘱的实体识别,提高临床医疗的工作效率。以上海市胸科医院为研究背景,提出嵌入于临床业务信息系统的临床医嘱实体识别方法。基于历年医嘱数据建立专项语料词库,运用CRF模型进行实时实体识别。随着识别系统上线使用,共处理了8362条医嘱,实验结果表明其准确率较好,在信息支撑度方面提升了医护人员的满意度。医嘱实体识别技术能够有效提高执行效率和医疗质量,也为医疗领域的知识挖掘工作提供了参考依据。Named entity recognition technology is applied to clinical business systems to realize the entity recognition of clinical medical orders and improve the efficiency of clinical medical treatment.Taking the Shanghai Chest Hospital as the research background,this paper proposes the entity recognition method of clinical medical orders to embed in the clinical business information system.Based on the data of medical orders over the years,a special corpus was established and the CRF model was used for real-time entity recognition.With the online use of the recognition system,a total of 8362 medical orders were processed.The experimental results show that the accuracy rate is good.And it improves the satisfaction of the medical staff in terms of information support.Entity recognition method of medical orders can effectively improve the execution efficiency and medical quality,and also provide a reference for knowledge mining in the medical field.

关 键 词：临床医嘱信息 CRF模型 命名实体识别 CPOE系统 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]