基于BiLSTM-CRF的中文临床指南治疗事件抽取

中文临床指南可以帮助医生在诊断、治疗疾病时做出恰当的决策。通过构建临床指南事件模型，利用词向量、长短期记忆网络、条件随机场等技术抽取中文临床指南中的治疗事件，为临床诊疗、用药等提供数据基础。相较于传统机器学习的抽取方法，本文建立的事件抽取方法具有准确率高、可移植性好的特点。

BiLSTM and CRF-based extraction of therapeutic events from Chinese clinical guidelines

Chinese clinical guidelines can help doctors to make an appropriate decision for the diagnosis and treatment of diseases. The therapeutic events were extracted from the Chinese clinical guidelines using word sectors, long- and short-term memory networks, and conditional random field in order to provide data for the clinical diagnosis and treatment of diseases and the use of drugs by establishing the model of events in Chinese clinical guidelines. The events extraction method established in this paper is characterized with a higher accuracy and a better transferability compared with that established on the basis of traditional machine learning.