Construction And Research Of Knowledge Map Of Stroke Medicine Based On Deep Learning

Along with the development of medical information and the general application of relevant technologies such as big data,the opening and sharing of medical big data,mining useful information in the data and realizing the wide application of data are of great significance to the development of The Times.Medical data contains a large amount of disease information.Extracting target information and mining potential links between information can provide rich data support for analyzing and exploring related factors of diseases.It is an excellent solution to solve and process medical data to construct medical knowledge graph by using knowledge graph technology and medical knowledge,to model knowledge relationship in medical information,to extract knowledge in medical information and store it.Therefore,this paper constructs knowledge map of stroke disease based on deep learning and knowledge map technology,mining potential information in medical records,in order to provide data support for the prevention and treatment of stroke disease.The main research contents of this paper are as follows:1.Construct data set of medical cases of stroke.By collecting the data of stroke medical cases in the TCM big data analysis platform of TCM think tank,and using a series of data preprocessing work such as data cleaning to further process the collected medical cases.On the basis of analyzing and pre-defining the types of entities and relationships in the medical record text,according to the characteristics of the knowledge map,this paper defines six entities and five relationships,constructs the entity and relationship extraction corpus,and marks the entities and relationships in the medical record through the doccano text annotation platform,completing the construction of the data set of the medical case of stroke.2.Construct knowledge extraction model.Aiming at the annotated data set,the BERTCas Rel information extraction model is constructed by using the joint extraction method to extract the triplet of < head entity,relation entity,tail entity > in the medical record text,and the model is trained by using the kaggle platform to realize the knowledge extraction in the medical record text.The comparison model was set up,and Precision,Recall and F1-Score were used as the evaluation criteria to evaluate and analyze the effect of model knowledge extraction.The comparative experimental results show that the model proposed in this study can effectively complete the entity relationship joint extraction task,which provides reliable data support for the construction of stroke knowledge map in the next step.3.Construct the knowledge map of stroke disease and realize the visualization of the knowledge map.Following the basic process of knowledge graph construction,after completing the triplet knowledge extraction of "head entity-relationship-tail entity" by using the knowledge extraction model,the six entities and five relationships extracted from the medical record text were converted into nodes and relationships in the Neo4 j graph database by writing python code,completing the creation of entities and relationships in the knowledge graph.The visualization and retrieval of knowledge map are realized.4.Taking the established knowledge map of stroke disease as a high-quality knowledge source,design and implement a disease and drug question answering system based on the knowledge map of stroke disease.The main function of the system is to convert the problem information into Cypher query language in the graph database by means of Q&A,and input the converted information into Neo4 j database to query the created knowledge map of stroke disease,so as to determine the stroke-related information corresponding to user input information.And give answers according to the query results of knowledge graph.In this paper,a comprehensive analysis of TCM medical cases of apoplexy was carried out,and it was used as the data source to construct a corpus of entity and relationship extraction,complete entity extraction and relationship extraction,and construct a knowledge map of apoplexy to systematically sort out the clinical symptoms and treatment methods of apoplexy,and design and implement a drug question and answer system of apoplexy disease based on the constructed knowledge map.The extended application of knowledge map has been realized,which has a positive effect on promoting the effective treatment of apoplexy.