Research And Implementation Of A Knowledge Graph Based Personalized Physical Examination Recommendation System

With the gradual improvement of living standards,many people have also increased their emphasis on their own health.Regular physical examination is an important means of assessing and monitoring physical health,which can effectively detect abnormal physical signs,and is an effective means of disease prevention and control.The fixed physical examination packages provided by various physical examination institutions and hospitals have many shortcomings,such as a wide variety of packages,excessive package items,etc.,which increases the difficulty of selection for physical examiners,and even causes excessive physical examination problems.The physical examination recommendation system considering individual differences has great potential to improve the deficiencies of the existing physical examination services.In order to recommend effective and targeted physical examination items,the intelligent physical examination recommendation system needs to include all kinds of information related to health and disease into the knowledge base of the recommendation system.Traditional recommendation systems are based on knowledge bases such as rule bases,electronic medical records,and health records,but these knowledge bases often have problems such as sparse data,and cold start.Knowledge graph(KG)can represent a large number of concepts and relationships in a clear and explicit way,and its association and reasoning ability can enrich and improve data.Thesis establishes a personalized medical examination recommendation system based on knowledge graph,which can customize medical examination items based on users’ personal information to meet personalized needs.The system has been tested and evaluated to analyze its effectiveness.The main work of thesis includes:1.Deeply analyze the data requirements of medical KG,fully utilize the relevant technology of KG to complete the extraction and fusion of multi-source heterogeneous data.Propose a method that combines knowledge representation with Euclidean distance calculation to mine implicit triplet information,and finally construct a medical KG.The medical knowledge graph constructed using efficient knowledge graph construction methods and knowledge graph completion methods based on reliable medical websites and medical data has reliability and completeness.2.Propose a personalized physical examination recommendation method based on medical KG and probability algorithms.This method can obtain the probability of suspected diseases based on user information,and then recommend physical examination items corresponding to suspected diseases.The validation results show that the average accuracy of disease prediction is 83.4%.3.Develop a personalized physical examination recommendation system.The main function of the system is to predict suspicious diseases for users according to their personal information,and then recommend targeted physical examination items,and recommend medical knowledge corresponding to suspicious diseases and physical examination items for users to help users understand disease information Assist users in physical examination.The main modules of the system include health questionnaire module,physical examination item recommendation module and medical knowledge recommendation module.The system was qualified through real machine debugging,cloud testing services,compatibility testing,and network testing,and the results showed that the system performed well and met the system standards.4.After implementing the personalized physical examination recommendation system,conduct systematic clinical trials.A certain sample size of electronic medical record data was selected from a tertiary hospital in China to verify the effectiveness of the system’s physical examination recommendation.Firstly,the electronic medical record data was organized into standardized data and imported into the system for testing results.Finally,the system was compared with standard results to evaluate its recommendation effectiveness.The results showed that the physical examination recommendation system had high recommendation accuracy,reaching 88.43%.At the same time,the appropriate method was selected to design the questionnaire of medical knowledge satisfaction,and the corresponding questionnaire survey activities were carried out.The final survey results showed that the medical knowledge recommendation effect of the physical examination recommendation system was good,with an average score of 4.43 points.