Research On Intelligent Medical Insurance Audit System Based On BP Neural Network And Association Rules

At present,the main function of the medical insurance audit system is still simple query,statistical analysis,multi-dimensional drilling and so on,providing very limited support for medical insurance regulatory work.With the increase of the number of insured persons and the diversification and concealment of the illegal forms,the medical insurance supervision led by the manual spot check is not effective,it can not discover medical insurance violations like the repeated user of drugs,repeat treatment and overdosing.The regulation of medical service behavior is directly related to the vital interests of the insured and the safety of the medical insurance fund,which influences the sustainable and healthy development of the medical insurance system.How to improve the informatization level of the current medical insurance audit system and provide decision support for the medical insurance supervision is an important way to solve the current loss of the medical insurance fund caused by the improper way of medical insurance supervision.By combing the related theories and the important algorithms of data mining and analyzing the application of data mining technology in medical field,the data mining technology is introduced into the intelligent medical insurance audit system.The data mining technology is used to deeply analyze,dig and model with the medical insurance data,and then apply the valuable knowledge extracted from data in the process of medical insurance supervision and managementso as to achieve full coverage,real-time and automatic inspection of the medical insurance information.As a result,providing a powerful means of information for medical insurance supervision.The main contents of this paper are as follows:Firstly,the flow,task and mining tools of data mining technology are teased,and BP neural network algorithm,association rule algorithm and attribute reduction algorithm are studied in detail.With the analysis of the applicability of data mining technology in the medical insurance audit system,an intelligent medical insurance audit system architecture based on BP neural network and association rules is proposed which including data acquisition layer,data preprocessing layer,data mining engine layer and service management layer.Secondly,the support library of intelligent medical insurance audit system support system is designed containing knowledge base,rule base and model base.In the design of the knowledge base,the process of constructing drugs interaction knowledge and disease common drugusing knowledge is mainly introduced.The main process of drug interaction knowledge construction is the extraction of terms,compositions and rules of interaction from drug specification corpus with the use of segmentation system.The disease common drug using knowledgeconstruction is based on the use of association rules algorithm,attribute reduction algorithm to analyse medical insurance prescription information so that found different disease common drug using model.The rule base covers the payment policy audit rule,the rationality audit rule of diagnosisand treatment,the clinical normative audit rule,the medical behavior abnormality monitoring rule and so on.The rule engine is a concrete implementation of rules,which converts the information of treatment into the data structure that the model can handle.In the model design,the structure of the BP neural network model including input layer,hidden layer,output layer and so on,as well as the model construction process are described in detail.Finally,integrating data mining tool WEKA into the development environment of Myeclipse10,and using the theory,method and technology of the research to develop the intelligent medical insurance audit system based on BP neural network and association rules,the system including knowledge base management,model base management,model application and other functional modules,and then illustrate the application of the system in the medical insurance supervision with the specific examples.