Study On Thinking And Logic Of Diagnosis And Treatment In Treatise On Febrile Diseases And Intelligent Diagnosis And Treatment Model Based On TCM State Identification

As the problem of aging population deepens in China,people’s needs for optimizing medical technology and enhancing health have become more urgent.In the new round of technological revolution and industrial transformation,artificial intelligence is an important driving force for all sectors of society.Traditional Chinese medicine(TCM)can only rejuvenate with the power of science and technology and its combinaton with cutting-edge artificial intelligence is inevitable in line with the trend of the times.Purpose:The purpose is to extract the traditional Chinese medicine thinking and modern logic of Treatise on Febrile Diseases(TFD),establish an electronic data set of standardized diagnosis and treatment of TCM with characteristics of TFD based on TCM state identification,and construct an intelligent auxiliary diagnosis and treatment model of TCM that reflects the characteristics of TFD to realize a computerized and intelligent recommendation from TCM symptom data to clinical medication.Methods:1.Conduct philology research through literature books,CNKI and other databases,and use "Treatise on Febrile Diseases" or "Treatise on Cold Pathogenic and Miscellaneous Diseases" or "Zhang Zhongjing" or "classical prescription" plus "thinking" as the main keywords of literature.Search and refer to relevant documents and materials from Baidu academic,health authorities,etc.,explore and sort out books and documents containing articles and prescriptions of TFD,and summarize the pattern of TCM diagnosis and treatment in TFD.2.Carry out in-depth study of the articles of TFD and further clarify its hidden modern logic of diagnosis and treatment through modern logic methods.3.Select all articles containing prescriptions in TFD through the research of state in TCM and optical character recognition technology and divide them into categories,further clarify the disease location and disease information in the articles,and then established the standardized electronic diagnosis and treatment data set for TFD.4.Construct a TCM diagnosis and treatment algorithm model based on the computer graph neural network algorithm(GNN)and multi-layer perceptron(MLP)technology,realize the computerized intelligent recommendation from TCM symptom data to clinical medication,and evaluate the performance of the model by three indicators: precision rate,recall rate and F1 score.Results:1.The TCM diagnosis and treatment thinking in TFD mainly includes the follwing aspects: Tangfang identification,pulse identification,symptom identification,dynamic identification and state identification.Specifically,Tangfang identification is the core diagnosis and treatment thinking.Pulse identification can be further divided into different situations: similar symptoms with different pulses,different symptoms with the same pulse,and one symptom with multiple symptoms.Symptom identification is divided into aspects including symptom connection,symptom comparison,and true and false symptoms.Dynamic identification is classified into pulse dynamics and phase.State identification is to identify the disease and symotoms displayed in different stages of the human body,and the location and nature of the disease are the final results of identification.2.TFD mainly contains definition logic,inductive and deductive logic,analysis and synthesis logic,as well as concrete and abstract logic.Various types of logic often appear in the same article,forming a holistic and inseparable part.3.A total of 358 data sets of diagnosis and treatment in TFD have been built,including1407 TCM symptoms data,416 TCM state data,1823 state factors(the location and nature of the disease)data,and 1846 TCM medicine data.There is a total of 5492 pieces of information from 4 dimensions.4.A TCM intelligent diagnosis and treatment model based on multi-graph neural network(MGNN)is constructed,which mainly covers TCM data feature aggregation module and TCM medicine prediction module.The optimal number of hidden layers of the GNN module is set to 800,and the optimal number of hidden layers for the MLP module is set to 1000.For the feature extraction module,the optimal N value in the Se graph is set to 3,and the optimal number of convolutional layers in the graph is set to 1.5.The performance of the intelligent diagnosis and treatment model of TCM based on MGNN is the best.The precision rate is 63.04% when the K value is 5,39.04% when the K value is 10;and the recall rate is 74.10%,when the K value is 5,87.05% when the K value is10;the F1 score is 68.07% when the K value is 5,53.84% when the K value is 10.Conclusion:1.Summarize the TCM diagnosis and treatment thinking in TFD through literature research,covering Tangfang identification,pulse identification,symptom identification,dynamic identification,and state identifiction.The five types of identification all embody the universal TCM diagnosis and treatment rule in TFD where the six meridians are used to find the location of the disease and eigh-principle pattern identification is used to determine the nature of the disease.2.Sort out the modern logics in TFD through modern logic research,covering definition logic,inductive and deductive logic,analysis and synthesis logic,and concrete and abstract logic.3.A standardized electronic diagnosis and treatment data set of TFD is established through TCM state identification method.4.A TCM intelligent diagnosis and treatment model based on MGNN is constructed through computer algorithms,which mainly covers TCM data feature aggregation module and TCM prediction module.5.The performance of TCM intelligent diagnosis and treatment model based on MGNN is better than that of traditional TCM diagnosis and treatment model through the comparison between model evaluation indicators.