Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (21): 39-51.DOI: 10.3778/j.issn.1002-8331.2303-0237

• Research Hotspots and Reviews • Previous Articles     Next Articles

Advances in Named Entity Recognition in Electronic Medical Record

LIU Andong, PENG Lin, YE Qing, DU Jianqiang, CHENG Chunlei, ZHA Qinglin   

  1. 1.School of Computer Science, Jiangxi University of Chinese Medicine, Nanchang 330004, China
    2.Second Affiliated Hospital, Jiangxi University of Chinese Medicine, Nanchang 330004, China
  • Online:2023-11-01 Published:2023-11-01



  1. 1.江西中医药大学 计算机学院,南昌 330004
    2.江西中医药大学 第二附属医院,南昌 330004

Abstract: Electronic medical record named entity recognition(NER) aims to identify medical entities in electronic medical record texts, and classify them into predefined medical entity categories. It provides support for further natural language processing tasks, such as medical relationship extraction, medical information retrieval, and medical intelligent question answering, etc. Firstly, the definition, labeling methods, evaluation indicators and difficulties of named entity recognition in electronic medical records are systematically sorted out. Secondly, the advantages of each type of named entity recognition methods in electronic medical records are summarized from two perspectives:the difficulty of named entity recognition in electronic medical records and the technology development process and deficiencies. Then, the evaluation tasks and data sets of named entity recognition in the domestic medical field are sorted out in detail. Next, solutions to each type of difficulty in electronic medical record named entity recognition are discussed and summarized in detail. Finally, the full text is summarized and the medical field is prospected the development direction of named entity recognition.

Key words: natural language processing, electronic medical record, named entity recognition

摘要: 电子病历命名实体识别(named entity recognition,NER)旨在识别电子病历文本中的医疗实体,并将其归为预定义的医疗实体类别,为进一步的医疗关系抽取、医疗信息检索、医疗智能问答等自然语言处理任务提供支持。系统梳理了电子病历命名实体识别的定义、标注方法、评价指标及难点;从电子病历命名实体识别难点及技术发展历程两个角度,综述了每类电子病历命名实体识别方法的优势与不足;详细梳理了国内医疗领域命名实体识别的评测任务及数据集;详细讨论和总结电子病历命名实体识别每一类难点的解决方案;总结全文并展望了医疗领域命名实体识别的发展方向。

关键词: 自然语言处理, 电子病历, 命名实体识别