Researches And Applications Of Knowledge Graph And Link Prediction Model For Famous Prescriptions Of Traditional Chinese Medicine

In recent years,the inheritance and development of traditional Chinese medicine(TCM)has been vigorously promoted.With the growing of the aging population,people’s health awareness has generally increased.The concept of early intervention,i.e."preventing diseases",in traditional Chinese medicine may play an important role in disease prevention.However,lots of knowledge in traditional Chinese medicine is disorganized or widely used with various forms,and it is often seen that the same knowledge is described in different terms.Therefore,it is truly necessary and significant to fully integrate the "multi-source and heterogeneous" TCM knowledge,well organize,manage knowledge and mine hidden TCM knowledge for the inheritance,development and applicaition of TCM.Nowadays,artificial intelligence has exhibited competitive capabilities in knowledge organization,management,association and visualization,and so on.Adopting artificial intelligence in TCM not only can solve many intractable problems,but also improve the application of TCM knowledge in Chinese herbal prescription improvement,drug relocation,drug discovery,etc.This will dramatically promote TCM impacts in National Health Service in the future.In the era of big data,the knowledge graph,one of artificial intelligent technologies,can effectively solve the above problems.As a knowledge network,a knowledge graph can contain rich semantic information,which can connect and represent scattered and complex knowledges.Because of its great performance in automatic question answering,knowledge retrieval,etc.,it has been widely used in many industrial fields.At present,more and more attention has been paid to knowledge graphs in the medical applications,and it has become one of the most important research domains of knowledge graphs.However,the research on the construction of knowledge graphs in traditional Chinese medicine is still in its infancy,and important hidden knowledge is not fully explored.Therefore,building a knowledge graph for TCM famous prescriptions can effectively manage related TCM entities such as TCM herbal prescriptions,Chinese herbs,diseases,and efficacies.Constructing an accurate and specific knowledge graph for TCM famous prescriptions provides an important foundation for mining hidden knowledge.This article describes the progress of the knowledge graph construction technology and the application of the knowledge graph in traditional Chinese medicine,constructing the knowledge graph for TCM famous prescriptions,and its drug relocation application.The details of this work include the following: First,this article has completed a knowledge graph for TCM famous prescriptions.Therein,the entities in the knowledge graph are defined,which mainly include four entities of "drug","efficacy","composition",and "disease".Then,the semantic relationship between the four entities is defined according to the characteristics of the entities.Finally,the knowledge is associated,integrated and visualized.Second,aiming at the task of named entity recognition(NER)related to famous Chinese medicine entities,here we design a deep learning-based BERT+Bi LSTM+CRF TCM NER model,and its highest F1 value reaches 0.9743,achieving a good entity Recognition performance.Third,a knowledge graph link prediction model based on deep learning is proposed.The proposed knowledge graph link prediction model,SK-Conv E,which integrates SKNet module.The resulting model is compared with the reported four representative models.The experimental results show that the prediction ability of the SK-Conv E model is improved compared with other tested models,and its Hit@1 value reaches0.234,which exihits a better performance in link prediction tasks than compared models.Fourth,the SK-Conv E link prediction model is applied for the constructed knowledge graph of TCM famous prescriptions to mine hidden knowledge,relocate drugs,and give an example of drug relocation taking the pandemic "Covid-19" "Infertility","Vaginitis","Lung cancer" as an example.Relocate drugs to traditional diseases to prove the feasibility of the results,For the new type of disease " Covid-19",the results show that the three prescriptions of Xiefeisan,Huanglian Jiedu Decoction,and Puji Disinfection Drink are the most likely to have inhibitory effects on the new coronavirus.