微型电脑应用

为了构建更加完善的医院智能分诊系统，替代大量的人工分诊，提高医疗服务效率，提出用于深度学习的智能科室分诊系统所需数据的自动标注方法和训练科室分诊模型。通过采集付费问诊的用户行为数据，解决训练需要的大数据集问题，并制定低成本的标注方案，构建出能够较好地满足智能分诊系统需要的大量标注数据；构建基于双向LSTM神经网络的深度学习模型。试验结果表明，该模型取得了较高的分诊准确度，总体达到一般医生的分诊水平。

In order to construct a more all-round hospital intelligent triage system which can replace human effort and improve efficiency of medical services, this paper aims to solve the problem of the construction of the labeling data required by an intelligent triage system based on deep learning. By collecting the user behavior data of paid consultation, we solve the problem of high-standard big data set which satisfies training requires, build a low-cost labeling scheme, and construct a large amount of labeling data which can meet the needs of intelligent triage system. At the same time, by constructing a deep learning model based on bidirectional LSTM neural network, a higher degree of accuracy of the department triage model is obtained, which generally reaches the diagnostic level of doctors.