ALBERT预训练模型在医疗文书命名实体识别中的应用研究
The Application Research of Named Entity Recognition in Medical Document Based on ALBERT Pre-Training Model
庞秋奔 1李银2
作者信息
- 1. 广西医科大学第一附属医院信息中心,广西南宁 530021
- 2. 中国移动通信集团广西有限公司网络运营中心,广西南宁 530012
- 折叠
摘要
中文电子病历命名实体识别主要是研究电子病历病程记录文书数据集,文章提出对医疗手术麻醉文书数据集进行命名实体识别的研究.利用轻量级来自Transformer的双向编码器表示(A Lite Bidirectional Encoder Representation from Transformers,ALBERT)预训练模型微调数据集和 Tranfomers 中的 trainer 训练器训练模型的方法,实现在医疗手术麻醉文书上识别手术麻醉事件命名实体与获取复杂麻醉医疗质量控制指标值.文章为医疗手术麻醉文书命名实体识别提供了可借鉴的思路,并且为计算复杂麻醉医疗质量控制指标值提供了一种新的解决方案.
Abstract
The main focus of named entity recognition on Chinese electronic health record is the study of progress note documentation datasets in electronic health record.This paper proposes a study on named entity recognition of medical surgical anesthesia document datasets.By leveraging the A Lite Bidirectional Encoder Representation from Transformers(ALBERT)pre-trained model to fine-tune the dataset and utilizing the trainer in transformers to train the model,the article realizes the recognition of named entities for surgical anesthesia events and the acquisition of complex anesthesia medical quality control index values on medical surgical anesthesia documents.This research provides a reference idea for named entity recognition in surgical anesthesia documents,and a new solution for calculating complex anesthesia medical quality control index values.
关键词
命名实体识别/轻量级来自Transformer的双向编码器表示(ALBERT)模型/transformers/麻醉医疗质量控制指标/医疗手术麻醉文书Key words
named entity recognition/A Lite Bidirectional Encoder Representation from Transformers(ALBERT)model/transformers/anesthesia medical quality control index values/surgical anesthesia documents引用本文复制引用
出版年
2024
NETL
NSTL
万方数据