International Journal of Medicine and Data
International Journal of Medicine and Data. 2023; 7: (6) ; 10.12208/j.ijmd.20230095 .
总浏览量: 540
贵州医科大学 贵州贵阳
*通讯作者： 莫大利,单位：贵州医科大学 贵州贵阳；
电子病历包含患者的病史、检查结果、诊断和治疗方案等重要医学信息，是支撑医疗智能化研究不可或缺的数据来源。然而由于电子病历的语言表达多样且复杂，医务人员往往难以快速准确地进行信息抽取和分析。基于深度学习的命名实体识别已成为电子病历自动化信息抽取的核心技术。但是医学自然语言处理的复杂性和医疗实体独特性导致电子病历的命名实体识别仍然存在挑战。本文通过构建BERT-MNER模型训练数据集和利用BERT预训练语言模型获取每个字的上下文相关嵌入表示，结合BiLSTM和CRF层完成医疗实体识别。实验结果表明，相较于基于单一深度学习网络的命名实体模型，BERT-MNER模型能更有效地识别电子病历中的医疗实体。
Electronic medical record contains important medical information such as patients' medical history, examination results, diagnosis and treatment plans, and is an indispensable data source to support intelligent medical research. However, due to the diverse and complex linguistic representations of electronic medical records, it is often difficult for medical professionals to extract and analyze information quickly and accurately. Named entity recognition based on deep learning has become a core technology for automated information extraction from electronic medical records. However, the complexity of medical natural language processing and the uniqueness of medical entities lead to the challenges of named entity recognition in electronic medical records. In this paper, medical entity recognition is accomplished by constructing a BERT-MNER model training dataset and using a BERT pre-trained language model to obtain contextually relevant embedding representations of each word, combined with BiLSTM and CRF layers. The experimental results show that the BERT-MNER model can identify medical entities in electronic medical records more effectively than the named entity model based on a single deep learning network.
[1] Zhang, R., Zhao, P., Guo, W., Wang, R., & Lu, W. (2022). Medical named entity recognition based on dilated convolutional neural network. Cognitive Robotics, 2, 13-20.
[2] Govindarajan, S., Mustafa, M. A., Kiyosov, S., Duong, N. D., Raju, M. N., & Gola, K. K. (2023). An optimization based feature extraction and machine learning techniques for named entity identification. Optik, 272, 170348.
[3] Wang, X., Liu, R., Yang, J., Chen, R., Ling, Z., Yang, P., & Zhang, K. (2022, May). Cyber threat intelligence entity extraction based on deep learning and field knowledge engineering. In 2022 IEEE 25th International Conference on Computer Supported Cooperative Work in Design (CSCWD) (pp. 406-413). IEEE.
[4] An, Y., Xia, X., Chen, X., Wu, F. X., & Wang, J. (2022). Chinese clinical named entity recognition via multi-head self-attention based BiLSTM-CRF. Artificial Intelligence in Medicine, 127, 102282.
[5] Rezayi, S., Dai, H., Liu, Z., Wu, Z., Hebbar, A., Burns, A. H. & Li, X. (2022, September). Clinicalradiobert: Knowledge-infused few shot learning for clinical notes named entity recognition. In International Workshop on Machine Learning in Medical Imaging (pp. 269-278). Cham: Springer Nature Switzerland.
[6] Xiong, Y., Peng, H., Xiang, Y., Wong, K. C., Chen, Q., Yan, J., & Tang, B. (2022). Leveraging Multi-source knowledge for Chinese clinical named entity recognition via relational graph convolutional network. Journal of Biomedical Informatics, 128, 104035.
[7] Li, J., Wei, Q., Ghiasvand, O., Chen, M., Lobanov, V., Weng, C., & Xu, H. (2022). A comparative study of pre-trained language models for named entity recognition in clinical trial eligibility criteria from multiple corpora. BMC medical informatics and decision making, 22(3), 1-10.