Construction Of Knowledge Graph Of TCM Prescriptions For Cardiovascular Diseases Based On Deep Learning

With the intensification of the aging of the population in modern society,the prevalence rate of cardiovascular disease is also increasing year by year,and showing a younger trend.There is no corresponding disease name in the field of traditional Chinese medicine,but according to the records of relevant syndromes and corresponding treatments,it belongs to the categories of "palpitation","chest arthralgia","vertigo" in the field of traditional Chinese medicine.Traditional Chinese medicine has been used to treat cardiovascular diseases by promoting blood circulation and removing blood stasis,regulating qi and other methods,and a good therapeutic effect has been obtained.With the rapid development of artificial intelligence,the combination of medical field and artificial intelligence has rapidly become a research hotspot,and has made important achievements in many aspects,such as knowledge question and answer,clinical decision-making and so on.Traditional Chinese medicine,as an indispensable part of the medical field,plays a vital role in the development of medicine.However,there are some problems in the field of traditional Chinese medicine knowledge,such as different sources of knowledge,uneven quality,trivial knowledge and so on,which not only increases the difficulty of the application of traditional Chinese medicine knowledge,but also brings certain challenges to the inheritance and development of traditional Chinese medicine knowledge.Knowledge graph technology visually displays complex knowledge in the field of traditional Chinese medicine through a graphical way,and integrates trivial information to help us find the hidden relationship between prescriptions and drugs,diseases and so on.it has contributed innovative methods to solve the problems in the field of traditional Chinese medicine.In order to solve the problems of boundary ambiguity and ambiguity in the process of named entity recognition in the field of traditional Chinese medicine prescription,this paper proposes a named entity recognition method based on attention mechanism bi-directional coding model(BERT),bi-directional long-term and short-term memory network(Bi-LSTM)and conditional random field(CRF).Knowledge fusion is carried out for the data with multi-source heterogeneity from different sources,and the methods of entity unification and entity disambiguation are used to make the data more clear and complete;the data are stored in Neo4 j diagram database to realize the visual display of TCM prescription knowledge graph;finally,the knowledge graph is completed based on the link prediction model to supplement the missing entity information and the relationship between entities.The main contents of this study are as follows:(1)This study completes the task of named entity recognition of TCM prescriptions for cardiovascular diseases based on BERT-Bi LSTM-CRF model.First,the BERT module is pre-trained to receive the corresponding word vector,and then the bi-directional long-term and short-term memory network(Bi-LSTM)module is input to capture the semantic information of the text context.Finally,the conditional random field(CRF)module is used to decode the output of the predictive label sorting,in order to complete the identification of prescription,drug,disease,efficacy and processing methods.This method has high applicability to the recognition of all kinds of entities in traditional Chinese medicine prescriptions,and the recognition accuracy of various entities in traditional Chinese medicine prescriptions has also been significantly improved.(2)The data of traditional Chinese medicine prescriptions for cardiovascular diseases were fused through the task of knowledge fusion.In view of the synonyms and synonyms that often occur in the data of traditional Chinese medicine prescriptions for cardiovascular diseases,this study uses the methods of entity unity and entity disambiguation to fuse the data.The data after knowledge fusion is clearer and more standardized,which lays the foundation for the construction of the knowledge graph of TCM prescriptions.(3)To complete the construction and visual display of the knowledge graph of TCM prescriptions for cardiovascular disease.The TCM prescription data of cardiovascular disease completed knowledge extraction and knowledge fusion are imported into the map database,and the relational triple data such as(prescription,attending,disease)are stored by using the Neo4 j diagram database,and the powerful storage,retrieval and processing functions of the graph database are used to visually display the materialized information,functional knowledge and the relationship between entities and entities.At the same time,the construction of knowledge graph is completed.(4)Complete the task of completing the knowledge graph of TCM prescriptions for cardiovascular diseases based on the link prediction model.In this study,by comparing the convolution embedding model(Conv E)with other link prediction models,it is found that the Conv E model can not only contact the context information,but also enhance the interaction ability of the model and improve the prediction effect by 3% to 5%.Based on this,we can effectively complete the task of completing the knowledge graph,and supplement the missing information in the knowledge graph of TCM prescriptions for cardiovascular disease.Through the above research contents,this study completed the construction of the knowledge graph of traditional Chinese medicine prescriptions for cardiovascular disease.combined with deep learning technology,it is analyzed in detail from the aspects of named entity recognition,knowledge fusion,visual display of knowledge graph and link prediction.Through the knowledge graph technology,the knowledge in the field of traditional Chinese medicine can be deeply excavated and applied,and the relationship between traditional Chinese medicine can be shown more comprehensively and visually.At the same time,it also provides an effective reference for the application of knowledge retrieval,recommendation and clinical decision-making in the field of traditional Chinese medicine.