Construction of Knowledge Graph of Spleen and Stomach Diseases in Traditional Chinese Medicine Based on Neo4j

Traditional Chinese Medicine (TCM) has a lot of valuable experience in the treatment of spleen and stomach diseases. Over the years, a large number of clinical resources have been accumulated in electronic medical records (EMR), and researchers have also published many related research papers. In order to make better use of these data resources, this paper introduces a process and method to build a knowledge graph of spleen and stomach diseases in TCM, and takes the EMR of TCM for spleen and stomach diseases and related literature inscriptions as data sources, and selects four commonly used named entity recognition (NER) models for comparative experiments of NER. Then relation extraction is executed. Finally, the neo4j graphical database is used to realize the construction and visualization of the knowledge graph of the spleen and stomach diseases of TCM. The experimental results showed that the BERT-BiLSTM-CRF model is the best in extracting entities, extracting symptoms, Chinese and Western medical diagnoses, and other medical entities with an average F1 value of 88.31%, and extracting relationships between medical entities using the BiGRU-Attention model with an Acc value of 88.26%. The knowledge graph of spleen and stomach diseases constructed can be retrieved under neo4j from a number of diseases, symptoms, and others. The knowledge graph of spleen and stomach diseases can be retrieved from multiple perspectives, such as disease and symptoms under neo4j. The knowledge graph constructed in this paper will help to deeply explore the potential knowledge and intrinsic relationships in the clinical treatment data resources of spleen and stomach diseases of TCM, help to better inherit the experience, and improve the clinical level of TCM.