基于深度学习的中文病历病史智能分类研究

Intelligent Classification of Medical History in Chinese Medical Records Based on Deep Learning

自然语言处理(NaturalLanguageProcessing,NLP)是计算机科学领域及人工智能领域的一个重要方向,结合机器学习的自然语言处理,能够有效地把非结构化的自然语言转换为结构化的数据。医院的电子病历主要用于临床,对于电子病历的数据,往往需要重新组织才能开展研究,论文主要研究基于卷积神经网络(ConvolutionalNeural Network,CNN)、长短期记忆网络(longShort-TermMenmory,LSTM)等算法对中医病历资料进行病史信息字段自动分类与抽取,旨在解决医学病历混杂的文本信息中自动抽取所有病史信息的分类问题。总结实验结果发现,基于卷积神经网络(CNN)的病史信息分类抽取F1值为0.850 6,基于长短期记忆网络(LSTM)的病史信息分类抽取F1值为0.881 0,具有良好的分类效果。

Natural language processing(NLP)is an important direction in the field of computer science and artificial intelligence.Combined with natural language processing of machine learning,it can effectively transform unstructured natural language into structured data.Hospital electronic medical records is mainly used in clinical,the electronic medical record data,often need to organize to conduct research,this thesis mainly studies based on convolutional CNN,and LSTM algorithm for the classification of the medical records of the history information extraction,aimed at resolving medical records mixed automatic extraction of text information all history information classification problem.It was found that F1 was extracted with a value of 0.8506 based on the medical history information classification of convolutional CNN,and F1 was extracted with a value of 0.8810 based on the medical history information classification of LSTM.All of them have good classification effect.