Research And Design Of Real-Time Detection And Visualization System For Medical Insurance Fraud Based On Streaming Data

As an important part of social security,medical insurance has played a vital role in the health care of the whole people.However,since the implementation of the medical insurance medical system,fraudulent behavior of the medical insurance fund has continuously appeared,seriously infringing on the interests of others and hindering the long-term sustainable development of my country's medical insurance policy.Medical event streaming data has the characteristics of large data volume,rapid generation and frequent changes.How to discover medical violations in real time from large,fast and time-varying streaming data,the traditional processing method can no longer solve practical problems.Based on the research of real-time streaming data computing engine,this thesis has conducted an in-depth discussion on the realtime detection and recognition of medical event streaming data,scene visual monitoring and micro-service-based system architecture methods,designed and implemented real-time detection of medical insurance fraud based on streaming data The visualization system provides effective practice methods for real-time fraud detection and monitoring in the field of medical insurance.The main work of the thesis is as follows:(1)Processing of streaming data.Medical insurance data is streaming data continuously generated by thousands of medical events.In this thesis,the real-time calculation method of the streaming data framework is used to monitor and track medical insurance fraud,and a real-time data warehouse model is constructed based on the Kafka data bus to ensure the real-time transmission of streaming data.Providing a data source for the detection and processing of fraudulent violations.(2)Integration of detection algorithms.Fraud is not static,it is necessary to integrate detection algorithms conveniently to cope with changes in fraud.The thesis deeply analyzes the domain-oriented algorithm integration process.On the basis of stream data processing,it proposes a data set-based algorithm integration framework.The dynamic configuration of the integration structure completes the detection algorithm and data source,calculation result set and result display.The dynamic binding of the mapping relationship between the algorithm solves the problem of the adaptation of the algorithm data source and the result display data source,thereby ensuring the real-time detection of fraud.(3)Visualization of data.In order to quickly understand the meaning behind the data obtained by fraud detection,it is necessary to parse,transmit and visualize the detected data entities.The thesis determines the content of fraud detection.Based on the principle of interface isolation,a visual presentation model is constructed to decouple the data processing logic from the data presentation to achieve data reuse and versatility.Node programming is used to define the interaction between layers to complete the real-time display of fraud data and multi-dimensional deep drilling.(4)System overall design and key technology.Based on the characteristics of the dynamic change of fraud detection in the medical insurance field,based on the concept of software reuse,the overall system structure,technical architecture,development model,and functional integration architecture were designed.The system integration,application integration,and visual interaction services are designed in detail and the key technologies are implemented,mainly including realtime analysis and transmission processing of streaming data based on middleware monitoring,algorithm generalization application improvement adapted to the characteristics of streaming data and node programming.The technology completes the deep-drilling visual interactive operation divided in different levels and uses the microservice framework to complete the collaborative work response and monitoring between components.(5)Prototype system implementation.Based on the above work,the prototype system is implemented.The system operation results show the feasibility and effectiveness of the system architecture method.