Named Entity Recognition in Chinese Electronic Medical Records Based on ALBERT-IDCNN-CRF
Gongzheng Tang · 2022
The task of electronic medical record named entity recognition is an integral part of the medical informatics field, which lays a strong data foundation for medical research and medical practice. The main task of electronic medical records named entity recognition is to transform unstructured text in clinical electronic medical records into structured data, which is still facing challenges such as semantic deficiencies and inefficiencies. In this paper, we focus on Chinese medical electronic medical record text and propose a Chinese medical electronic medical record named entity recognition model based on fusion learning of the ALBert model, in which the ALBert pre-trained language model can better represent the contextual semantics in electronic medical record sentences, and the Iterated Dilated Convolutional Neural Network (IDCNN) has better recognition in convolutional coding of local entities and finally the sequence tagging of named entities based on the Conditional Random Fields(CRF). Experimental results on the standard dataset show that the method improves the F1 value by 7.03 % compared to the baseline model and significantly reduces the time overhead.