Named entity recognition method of Chinese EMR based on BERT-BiLSTM-CRF

Abstract

With the widespread use of Chinese electronic medical records, the extraction of medical named entities and the resolution of the polysemous term of entities are of great significance to the analysis and processing of patient information and disease diagnosis. This paper uses a BERT Chinese pre-training vector that does not rely on manual feature selection, combines BiLSTM and CRF Chinese named entity recognition algorithm model, and applies it to the processing of the CCKS2020 electronic medical record data set. This paper conducts experimental tests and comparisons between the BERT-BiLSTM-CRF model and other models. The results show that the results of this model are better than other models, and it can accurately identify multiple entity categories and has a good application prospect.