Realization Of Intelligent Prescription Auxiliary Recommendation System Based On Knowledge Graph

Traditional Chinese medicine(TCM)is a precious cultural heritage in China,and has played an important role in the medical field for thousands of years.However,traditional TCM is facing problems such as complex knowledge systems and difficult inheritance in the modern development.In recent years,the booming technologies such as artificial intelligence and knowledge atlas are powerful tools to assist the development and inheritance of traditional Chinese medicine.Based on the knowledge atlas of traditional Chinese medicine prescriptions,an intelligent recommendation system for traditional Chinese medicine prescriptions is developed,which can utilize classic Chinese medicine literature and modern scientific research achievements to inherit the essence of traditional Chinese medicine,and can combine traditional Chinese medicine theory with modern information technology,which is conducive to studying traditional Chinese medicine theory from multiple perspectives.Based on this background,relying on the data provided by Chongqing Bishan Hospital of Traditional Chinese Medicine and the guidance of traditional Chinese medicine experts,this thesis conducts a series of research on technologies such as the construction of traditional Chinese medicine knowledge atlas and recommendation of traditional Chinese medicine prescriptions,with the goal of improving the accuracy and comprehensiveness of prescription recommendations.Finally,it builds an intelligent prescription recommendation system with application value to improve the efficiency of doctors’ diagnosis and treatment.The main research work of this thesis is as follows:(1)Construction of TCM knowledge graph.Based on traditional Chinese medicine books such as the "Dictionary of Traditional Chinese Medicine Prescriptions" and "Chinese Materia Medica",multidimensional data such as prescriptions,herbs,and symptoms needed to construct a knowledge graph of traditional Chinese medicine are obtained through data processing methods.Data cleaning is then carried out,and a knowledge graph is developed to analyze the cleaned drug symptom data,determine the definitions of entities,attributes,and relationships in the knowledge graph,and then focus on the characteristics of different data,Use tools such as deep learning to define semi-automatic knowledge extraction algorithm to complete the construction of triple;Standardize the constructed knowledge graph by designing entity alignment rules and standardized representation of knowledge graph,and then store them in the Neo4 j graph database;Finally,visualization technology is used to visualize and render the established TCM knowledge graph,providing user interaction functions to help users understand TCM knowledge more systematically and comprehensively.(2)Implement a prescription recommendation algorithm that integrates knowledge graph and personalized diagnosis and treatment.Based on the traditional Chinese medicine knowledge graph constructed in this thesis,we propose a new multi task joint learning model SM-MKR,which can effectively utilize knowledge graph information to assist prescription recommendation tasks.Adopting the alternating learning concept in the knowledge graph recommendation system,the model is divided into knowledge representation learning tasks and prescription recommendation tasks;Designed an information dissemination unit for two modules to share training information;In the recommendation task,a factor decomposition machine based on attention mechanism is used to fuse personalized diagnosis and treatment features of patients,capturing high-order nonlinear features during the training process;Integrating recommendation task learning with the topological structure of knowledge graph through multi task alternating learning to improve the efficiency of knowledge representation task learning and further improve the accuracy of prescription recommendation tasks;Finally,the effectiveness was verified on a public dataset,and the experimental results showed that the SM-MKR model can effectively integrate multi-source heterogeneous information such as knowledge graphs and patient personal attributes,completing the auxiliary recommendation task of traditional Chinese medicine prescriptions.(3)Design and implement a multifunctional prescription recommendation system.The front and rear end separation mode is adopted,and development technologies such as Springboot and Python are used.In terms of database,relational database My SQL and graph database Neo4 j are used;Develop an intelligent auxiliary prescription system with multiple functions such as login registration,intelligent prescription,data analysis,and information retrieval;Finally,in order to ensure the stability and availability of the system,system functional testing and performance testing are conducted.