A Study On The Combination Law Of Four-Diagnosis Of Lung Cancer Medical Cases Based On Knowledge Graph

Lung cancer is the malignant tumor with the highest morbidity and mortality in China,which not only brings huge economic burden and living pressure to the patients’families,but also poses a serious threat to the patients’physical and mental health.With the strong support of the state and the rapid development of traditional Chinese medicine and other related fields,the characteristic therapy of traditional Chinese medicine has become one of the important methods for the treatment of cancer in China.However,with the development of medical information,the medical data are increasing day by day.These data are confused and complicated and contain rich semantic information,how to connect these scattered and trivial information to realize knowledge integration has become a hot issue in the field of medical research.In view of the existing problem,an effective solution is to build a medical-related knowledge graph.Knowledge graph is a network-based method to capture the scattered and disordered entities,concepts and their relations in various fields and present them more accurately and directly in a graphic way.At present,there are many mature large-scale knowledge bases in the market,but because of its own characteristics,the construction of Knowledge Atlas is still in the exploratory stage,which is not conducive to the dissemination and sharing of traditional Chinese medicine.Therefore,in order to mine the potential knowledge from the lung cancer medical record data,this paper uses the knowledge graph technology to construct the four-diagnosis information related of the lung cancer medical record data,it provides the research mentality and the theory support for the medical worker to carry on the lung cancer related topic exploration.The main work of this thesis is as follows:(1)In order to obtain abundant four diagnosis information from lung cancer medical records,this paper uses BiGRU-CRF method based on word vectorto extract related entities of four diagnosis information from lung cancer medical records.Firstly,the lung cancer clinical data automatically annotated based on the user-defined dictionary is pre-trained by the BERT model to get the word vector containing context semantics,and then it is input as BiGRU-CRF model to realize the named entity extraction of the four diagnosis information of lung cancer medical records.The experimental results show that the accuracy of this method is better than other models in terms of symptoms,tongue,pulse or degree adverbs,which indicates that this method has stronger ability of named entity recognition in the research of named entity extraction of TCM medical records,and can be better applied to the research of named entity extraction of TCM medical records.(2)In order to realize the analysis and mining of symptom groups and their combi-nation rules of the four diagnosis information of lung cancer medical records,this paper first transforms the lung cancer data into five co-occurrence matrices,and then uses Fast Unfolding algorithm to divide the communities.The results show that "clinical-clinical","clinical-tongue","tongue-pulse","tongue-tongue" and "clinical-pulse" can be divided into 5 groups,4 groups,3 groups,2 groups respectively.In this paper,the characteristics of symptom groups and their combination rules of the four diagnostic information of lung cancer medical records are preliminarily studied,which provides an experimental basis for the final construction of the knowledge graph of the four diagnostic information of lung cancer medical records.(3)In order to realize the construction and visual display of the knowledge graph related to the information of the four diagnostics of lung cancer medical cases,the Neo4j technology was used in this paper to transform the extracted entities and the relations between entities into a structured and graphical form of knowledge graph.The experiment can not only intuitively discover the relationship between the four diagnostic information entities,but also generate a variety of lung cancer medical information knowledge graphs according to the different needs of the experiment,which lays the foundation for further research in the field of lung cancer and the discovery of potential relationship between entities.