Research And Implementation Of Entity Relationship Extraction Based On Chinese EMR

Electronic medical record（EMR）is some clinical information of patients generated by medical staff in the medical process,including a large number of medical entities related to patients’ health.How to obtain medical knowledge in EMR through entity relationship extraction model is a problem worthy of attention and research.The current entity relationship extraction model is mainly a pipeline extraction model,which is a series design of two subtasks of entity recognition and relationship extraction.However,this method will be affected by redundant entity pairs,and can not well capture the internal relationship between entities and relationships.In view of the shortcomings of the pipeline model,this paper first implements a joint entity relationship extraction model based on the decomposition strategy,which adopts the strategy of extracting the head entity first,and then extracting the tail entity corresponding to the head entity for each relationship,which improves the interaction between the head and tail entities and solves the problem of entity redundancy.Then,according to the high-density distribution of Chinese EMR text entities and the cross interconnection of relationships between entities,a joint entity relationship extraction model for Chinese EMR（EMR-RE）is proposed.The model improves the decoder for identifying head entities and relationship specific tail entities by constructing multi-layer pointers,and improves the learning ability of the model in dealing with nested entities and entity overlapping triples in EMR text,Compared with the current mainstream electronic medical record text processing model Bert multihead,the F1 value of this method is increased by 2.1%.Then,by improving the category and the depth structure of the pre training model,the F1 value of the EMR-RE is increased by1.1%.Then,all the extracted triples are stored structurally,the medical knowledge base is constructed,and the basic information triples of patient EMR are visualized with the medical knowledge base to construct the knowledge map of patient medical record.Finally,by supplementing the medical knowledge base and using the AC multi pattern matching algorithm integrating similarity calculation,the medical decision-making auxiliary system is designed to realize the functions of patient medical record knowledge display and fine-grained Medical Knowledge Q ＆ A.