Named Entity Recognition in Electronic Medical Records Based on Bidirectional Long Short-Term Memory-Conditional Random Field

In most hospitals in China, the actions of medical staff, patients' conditions, and treatment methods for diseases all need to be recorded in a system for easy management and viewing. This system is called electronic medical record(EMR). Due to the large number of medical institutions in China, EMR contains a large amount of valuable information. The purpose of this study is to use deep learning to automatically identify key words in EMR. This study uses a Bi-directional Long Short-Term Memory (BiLSTM) network to extract sequence information, and uses a Conditional Random Field (CRF) to solve the label mapping problem. An open-source dataset is used to train and test the model. In the experiment, we collected multiple experimental data and evaluated the model's performance. The results show that the model can be well applied to the entity recognition task in EMR.