| At present,malignant tumor is a complex disease which seriously threatens human life and health.The advantages of traditional Chinese medicine in tumor prevention and treatment and postoperative conditioning are increasingly prominent.Lung cancer is one of the tumors with the high morbidity and mortality in China.Numbers of clinical practice and research have been suggested that lung cancer in the treatment of traditional Chinese medicine and integrated traditional Chinese and western medicine has achieved remarkable results.TCM master Zhou Zhongying has rich experience in the clinical treatment of tumors,and has accumulated numbers of clinical cases in the process of diagnosing and treating lung cancer patients,which is worthy to be learned by scholars and researchers.However,TCM medical records are the experience accumulation in the clinical diagnosis and treatment,which are manifested unstructured,scattered and individualized.And it is difficult and challenging to describe TCM knowledge uniformly and quickly obtain effective information from medical records.In addition,it is difficult for common statistical methods to completely reveal the knowledge contained in the medical cases of TCM masters.And learners cannot efficiently obtain and apply the experience of masters.So it is necessary to refine,summarize and learn the information of medical cases.In today’s explosive growth of knowledge,how to apply modern technology to study and inherit the academic thoughts and experience of TCM masters is an urgent problem to be solved in the field of TCM.Knowledge graph provides a new method for inheriting and carrying forward the academic thoughts and clinical experience of TCM masters.Objective:The construction of knowledge graph for diagnosis and treatment of Lung Cancer by TCM master Zhou Zhongying was studied based on the medical records of lung cancer,so as to comprehensively display the knowledge system and clinical path of professor Zhou Zhongying,realizing the experience display and retrieval of the TCM master.This study provides new ideas and methods for learning and inheriting academic thoughts and experience of Professor Zhou Zhongying in the treatment of lung cancer,broadens the ideas in the treatment of lung cancer,promotes the development of the academic experience of TCM masters,and lays a solid foundation for the application of TCM knowledge graph.Methods:(1)Systematically collect and organize the research status at home and abroad of knowledge graph application and TCM knowledge graph by using literature research method,and master the concept,construction process,and technology of knowledge graph,so as to learn ideas from existing research,and summarize the relevant research of knowledge graph construction in the field of TCM and the direction to be further improved,providing the theoretical basis and ideas for this study.(2)Collect the outpatient medical records in the treatment of lung cancer by professor Zhou Zhongying from July 1986 to February 2015.According to the inclusion criteria and exclusion criteria,446 cases of 576 records were screened as the research object.The content of medical records include patient information,clinical symptoms,four diagnostic information,pathogenesis,prescriptions,etc.The medical records were standardized and preprocessed.And the database of medical records on lung cancer was established.(3)Under the guidance of the seven-step method,Protege5.5.0 ontology tool was used to design and build the knowledge ontology framework of TCM master Zhou Zhongying in the treatment of lung cancer in reference to the list of TCM subject headings,TCM monograph textbooks,industry standards,etc.The framework comprehensively displays the pathogenesis,symptoms,Chinese herbs and other related concepts in lung cancer treatment,forming the pattern layer of knowledge graph,so as to provide support for mastering the knowledge system of Professor Zhou Zhongying and lay a foundation for the construction of knowledge graph.(4)Bert-BiLSTM-CRF model was used to identify symptoms,tongue and pulse in the medical records of lung cancer by Professor Zhou Zhongying.Combined with manual examination,the structured and standardized lung cancer entities were matched with the ontology framework of the pattern layer to form a data layer of knowledge graph.(5)The knowledge graph is stored in Neo4j database.The semantic query and visual display are realized through the database.Then,the evaluation of the knowledge graph was constructed by expert consultation to test the scientificity and validity of the knowledge graph.(6)Based on the above research,the data mining methods were used to analyze and summarize the relationship among pathogenesis,symptoms and prescriptions in the treatment of lung cancer by professor Zhou Zhongying,realizing the transformation of the implicit knowledge to explicit knowledge,which are added to the knowledge map to make the map more perfect and complete.Results:(1)Construction of ontology of diagnosis and treatment of lung cancer by TCM master Zhou Zhongying in TCM.Based on the medical records and academic ideas of professor Zhou Zhongying in the treatment of lung cancer,the knowledge ontology framework is constructed to define the entity properties and relationship between the entities.The knowledge ontology defines eight top-level concepts.including symptoms,pathogenesis,pathological nature,pathological location,pathological factors,cure rules,Chinese herbs,auxiliary check.A total of 34 classes,277 instances,7 object properties,5 data properties were extracted.It is tested that the ontology has internal logical consistency,and it was revised by experts from the inheritance studio of TCM master Zhou Zhongying.The framework was finally determined as the pattern layer of knowledge graph,which lays the foundation for the construction of knowledge base or knowledge graph.(2)Research on entity extraction of the medical records of lung cancer based on deep learning.Combined with the characteristics of Professor Zhou Zhongying’s lung cancer medical records,the Bert-BiLSTM-CRF model was used to identify the entities of symptom in the medical records of lung cancer.And three named entity categories of symptoms,pulses and tongues were identified.The accuracy rate of the model was 81.58%,the recall rate was 86.52%,and the F1 measure was 83.98%.The indicators are all above 80%,which suggest the overall effect of entity extraction is good.In combination with manual approval and entity alignment,the structured and standardized labeled lung cancer entity data is matched with the ontology framework of the pattern layer to form a data layer of the knowledge graph.(3)Storage and application of knowledge graph based on Neo4j graph database.The Neo4j database was used to build the TCM knowledge graph for the diagnosis and treatment of lung cancer by professor Zhou Zhongying.The data were imported into the Neo4j database through Python.The knowledge graph includes 1281 entities and 7554 relationships,which were used for the visual display and semantic retrieval function.And the quality evaluation of knowledge graph was carried out,and the results suggest the knowledge graph constructed in this study is comprehensive and reliable,the mode layer is reasonable,and it can comprehensively show the medical cases of lung cancer,which is scientific and practical.(4)Improvement and revision of the knowledge graph in the diagnosis and treatment of lung cancer by TCM master Zhou Zhongying.The data mining methods were applied to analyze the relationship among symptoms,herbs and pathogenesis in lung cancer treatment.The common pathogenesis,symptoms and medication of Professor Zhou Zhongying in diagnosis and treatment of lung cancer were summarized based on data mining.The main herbs associated with suspending fluid retention pathogenesis are Mori cortex,Descurainiae semen lepidii semen,Astragali radix,Pinelliae rhizome praeparatum,Perillae fructus,Benincasae exocarpium,Stephaniae tetrandrae radix.The main symptoms of suspending fluid retention pathogenesis are dyspnea,chest pain,side pain,chest tightness,cough,asthma and other symptoms.After the confirmation of the inheritance studio of TCM master Zhou Zhongying,it was added to the knowledge graph to improve and revise the existing knowledge graph.Conclusion:Knowledge graph is an emerging mode of knowledge management and service in the era of big data,it can capture and present the relationship between concepts of TCM in a visual way,laying the foundation for TCM knowledge service and providing new methods for inheriting the academic thoughts and clinical experience of TCM masters.Based on the lung cancer medical records of TCM master Zhou Zhongying,the knowledge ontology framework was designed.The TCM knowledge graph for treatment of lung cancer by professor Zhou Zhongying based on the framework was explored,evaluated and improved,which realizes TCM experience knowledge visualization and semantic retrieval functions,and provides methodological reference for inheritance and development of academic experience of TCM masters. |
