Research On Tcm Clinical Knowledge Graph And TCM Intelligent Application Technology

The core goal of medical artificial intelligence is to achieve clinical assisted diagnosis and treatment,thereby helping to improve clinical diagnosis and treatment capabilities.Because medical knowledge graph has rich domain knowledge and often plays a key role in clinical diagnosis and treatment,knowledge graph and its reasoning application research play an important role in medical artificial intelligence.The clinical diagnosis and treatment process of traditional Chinese medicine(TCM)is a deductive reasoning process,so it is important to carry out auxiliary diagnosis and treatment based on knowledge graphs.However,this research has the following problems:(1)the construction of knowledge graph is a systematic project.The medical knowledge graph of medical knowledge graph needs to be continuously expanded,and there are problems of incompleteness;(2)in the process of auxiliary diagnosis and treatment,the application of medical knowledge graph often lacks the correlation between theoretical knowledge and clinical cases;(3)the existing auxiliary diagnosis and treatment systems are mostly based on the method of data mining,which mine hidden knowledge from a large number of original cases,cannot effectively use the theoretical knowledge in Chinese medicine books,so it cannot fundamentally understand the process and actual connotation of seeing a doctor.In order to solve the above problems,this paper carries out research work from the following three aspects:(1)Aiming at the problem of incomplete medical knowledge graph,a knowledge graph completion method named Gate TD is proposed for the completion of medical knowledge graph.This method introduces GRU gating mechanism on the basis of tensor decomposition,which effectively captures the interaction between entities and relations,and enables relation vectors to have different vector representations under different headentity conditions.This chapter conducts knowledge graph completion experiments on four standard datasets.Compared with the baseline method,it achieves a significant performance improvement on FB15 k dataset.The experimental results show that Gate TD enhances knowledge sharing through GRU,which can make the relation vector learn more implicit information from the head entity.Applying this method to the completion of medical knowledge graph has great practical significance and academic value for the construction of medical knowledge graph and the reasoning based on medical knowledge graph.(2)Aiming at the problem of lacking the correlation between theoretical knowledge and clinical cases,an analysis method of clinical case diagnosis and treatment rules based on analysis and learning is proposed.First,the connection between theoretical knowledge(symptoms,TCM)and clinical data(symptoms,TCM)in the knowledge graph is formed through entity alignment.It establishes a reliable association between theoretical knowledge in the knowledge graph and clinical data according to the screening results.This method has high reliability and has great practical significance for the construction of domain data and the discovery of diagnosis and treatment rules,and at the same time provides a basis for TCM clinical decision support.(3)Aiming at the problem that the TCM auxiliary diagnosis and treatment system cannot effectively utilize theoretical knowledge,a prescription recommendation method based on multi-strategy learning is proposed,which is based on the TCM knowledge graph and clinical data to achieve TCM clinical auxiliary decision support.Compared with machine learning multi-label classification methods,network mapping methods and baseline knowledge graph completion methods,the multi-strategy learning prescription recommendation method achieves the best performance,with Precision@10=0.259,Recall@10=0.238,F1@10=0.248.This result fully demonstrates that the knowledge graph provides a potential source of information for the prescription recommendation process due to its rich internal semantic associations,making the recommendation results more accurate,diverse,and interpretable.Therefore,knowledge graph has great application value and academic significance in the field of auxiliary diagnosis and treatment.