Research On Medical Insurance Abnormality Analysis And Judgment Method Based On Big Data

With the advancement of the new medical reform,China’s social medical insurance has developed rapidly,and the level of medical informatization has also been continuously improved.However,with the further deepening of the reform of the medical insurance system,the gradual increase of the number of insured persons and the gradual deepening of informatization,many unreasonable medical reimbursement behaviors have appeared in recent years.These behaviors violate the laws and regulations related to national medical insurance,and cheat a large number of medical insurance funds from national medical insurance institutions,hospitals,community clinics,pharmacies and other medical industry institutions by making up the physical condition and concealing the real situation,which seriously endangers the perfection and perfection of China’s medical insurance system.At present,after years of medical informatization and the application of various medical information systems,hospitals and medical insurance management departments have accumulated a large number of medical treatment records and electronic data,covering the treatment time,treatment expenses and so on.However,the disunity and inefficient use of existing medical information standards are difficult to promote data sharing,data mapping and data fusion based on medical insurance research and analysis.Therefore,how to effectively research and analyze the medical insurance data set under the background of big data and mine the hidden abnormal behaviors is a very valuable research problem,which has important practical significance and value for promoting the standardization and structure of medical insurance data,improving the efficiency of medical insurance abnormal detection and mining quality,and assisting medical insurance research and judgment.This dissertation relies on the National Natural Science Foundation of China’s key project "Big Data Driven Smart Medical Health Management Innovation" with grant number of 71532002 and the National Natural Science Foundation of China’s general project "Medical Health Big Data Driven Clinical Misdiagnosis and Judgment Theory and Technology" with grant number of 62173025.Based on the comprehensive review of domestic and foreign theoretical and technical documents,this dissertation focuses on the related methods of medical insurance anomaly analysis and big data mining technology.Introduce medical information standardization,relationship extraction,machine learning,big data analysis and other methods to study medical insurance data structure,medical insurance record coding assisted by big data,medical insurance abnormal behavior mining and medical insurance active judgment methods.The research contents and results are as follows:(1)Implementation method of medical insurance record coding based on HadoopFirstly,this dissertation analyzes the new features and advantages of ICD-11 compared with the old coding method,expounds the application and important role of ICD coding in current medical big data,and analyzes the potential problems faced by popularizing and applying ICD-11 standard under the current situation.Then,the evaluation form of ICD-11 implementation was established,and the specific steps of ICD-11 implementation in different stages were put forward through three stages: before implementation,during implementation and after implementation.According to the three key stages of reporting medical record data,migration and mapping of old coding standards and implementation of ICD-11 automatic coding,this dissertation designs from the perspective of big data analysis technology,and puts forward cross-regional ICD-11 information service architecture,medical record coding conversion and mapping method based on Map Reduce model and medical record automatic coding method oriented to big data analysis,which solves the key problems in the implementation of ICD-11 and improves the efficiency and accuracy of medical record coding conversion and upgrading.(2)Mining abnormal behaviors of medical insurance aggregation based on frequent patternsBased on the new medical insurance record coding system,firstly,the challenges faced by the detection of abnormal behaviors of medical gathering are analyzed.In view of the complexity of abnormal behaviors of medical gathering,the related concepts are given,and a structured model of medical gathering behaviors and a matrix of medical behaviors are constructed.The association rule mining method is introduced to transform medical aggregation behavior mining into frequent pattern mining.Using big data analysis technology,combined with Map Reduce distributed computing model and the actual scene of medical insurance reimbursement,an improved distributed-based medical aggregation behavior mining method is proposed,which overcomes the shortcomings of association rule mining method in medical field,improves the efficiency and accuracy of high-dimensional frequent itemsets data detection,and realizes the effective identification of abnormal medical aggregation behavior by analyzing and verifying the real outpatient reimbursement data.(3)Medical insurance fraud marking method based on patient information weighted tree.Based on the concept of multi-tree and aiming at similar medical claim information of different patient types,a weighted multi-tree of patient information is constructed,in which the column values of patient information are regarded as nodes,and each complete path between the root node and the leaf node is regarded as a detailed information set of the corresponding patient type.The weights and thresholds of edges are calculated according to the relative frequency of information sets,and the medical insurance reimbursement information of patients is mapped with the patient information list containing real fraudulent information to detect fraudulent claims.Compared with the traditional mapping method,the mapping based on weighted tree of patient information reduces the number of false positives by providing effective fraud markers,and improves the efficiency of fraud detection.(4)Active research judgment method of medical insurance based on intelligent big dataOn the basis of research contents(1),(2)and(3),firstly,this dissertation analyzes the limitations of existing big data theories in medical and health related applications,focusing on the complex problems encountered in the field of medical and health big data analysis,and puts forward the definition and conceptual model of intelligent big data.This model can effectively utilize the relationship between big data sets,and build intelligence by integrating key technologies and methods such as structured processing of medical insurance big data sets,semantic analysis of data sets,mapping of machine learning methods,and exploratory visualization.With the support of Hadoop,Hbase,Spark and other big data components,the intelligent big data analysis model and the active decision-making framework of medical big data proposed in this dissertation are applied to the medical insurance research and judgment support scenario with practical significance,which can finally provide reference for the theory and technology needed by the development of medical information in China in the era of medical health big data..