Traditional Chinese Medicine Consultation System Based On Knowledge Graph

The combination of traditional Chinese medicine and artificial intelligence technology is an important development direction of modern Chinese medicine.With the development of artificial intelligence technology,people can conduct preliminary self-diagnosis of diseases and obtain some medical advice through online consultation.TCM consultation is different from general online consultation.In long-term practice,TCM has developed its unique set of dialectical methods.TCM obtains symptoms through the four diagnostic methods of TCM,comprehensively summarizes the patient’s symptoms through the thinking of dialectical treatment,infers TCM syndromes,and finally determines the corresponding treatment methods.The research on dialectical diagnosis of traditional Chinese medicine has achieved good results from the initial expert system,machine learning classification dialectical model to intelligent interrogation system based on deep learning.Technologies such as natural language processing,knowledge graphs,and deep learning have further promoted the development of TCM modernization.At present,online consultation mostly adopts the method of rules,adopts the method of pattern matching to collect disease information,and then obtains related medical advice information based on the established knowledge base.This rule-based method is fast and has high accuracy,but it is difficult to cover all problems with manually established rules,and it requires a lot of manpower to maintain and update.Emerging intelligent consultation models usually use knowledge graphs to build medical knowledge networks and build machine learning or deep learning models,but their interpretability is poor.At the same time,data construction in the field of traditional Chinese medicine is difficult and the accuracy rate is low.On the other hand,the existing intelligent consultation focuses on Western medicine diagnosis scenarios,and there are few TCM consultations combined with the dialectical theory of TCM.In view of the above situation,this thesis combines artificial intelligence technology with TCM dialectical diagnosis and treatment methods,designs an intelligent inquiry system based on knowledge graphs to provide users with TCM dialectical diagnosis and treatment services,and proposes a judgment method of TCM syndromes.The main work content can be summarized as the following points:1)Construct a knowledge graph of TCM diagnosis and treatment.Research the construction method of knowledge graph,the characteristics of Neo4 j graph database,various methods of information extraction,collect relevant diagnosis and treatment knowledge to extract entities and relationships,and generate symptom query sentences to complete the construction of knowledge graph.2)Study the classic model BERT-Bi LSTM-CRF currently used in named entity recognition tasks,Analyze the role of each layer of the model,construct a doctor-patient dialogue dataset suitable for TCM consultation scenarios,carry out symptom entity recognition model training,and verify the effect of the model in the symptom recognition task of this thesis through comparative experiments.3)A method of TCM syndrome reasoning is proposed.The four diagnosis symptoms obtained in the patient’s consultation are used as parameters,the syndromes are used as conclusions,and 10 types of dialectics such as the eight principles of traditional Chinese medicine and the three energizers are used as the reasoning principles of syndromes to establish logical expressions.Design a method of counting the symptom parameters and logical symbols in the expression to verify the symptoms.The method uses the number of occurrences of the symptom parameters as the weight of the symptom parameters,the number of logical "AND" in the logical expression is used as the length of the proof path of the expression,so as to determine the symptoms to be asked in the next step,obtain the value of the symptom parameter in the expression,and perform syndrome reasoning.4)Design and implementation of TCM consultation system.Based on the previous work on symptom recognition and syndrome reasoning,a applets for consultation and backstage knowledge management system were designed and implemented.By analyzing the requirements of the system,the architecture and function design of the consultation system are determined,and the system implementation of the applets is completed.The consultation applets interacts with the patient and recognizes the symptom entity in the patient’s answer as a symptom parameter to perform syndrome reasoning.When the reasoning information is insufficient,the symptom verification is carried out,and the patient is asked whether he has relevant symptoms,so as to control the dialogue process with the patient,complete the symptom parameters,finally obtain the syndrome,and output the diagnosis and treatment results.The backstage knowledge management system adopts B/S architecture design,and the administrator can update the knowledge graph and manage other functions through the system.