Research On Risk Prevention And Control Algorithms For Medical Insurance Data

With the continuous improvement of the medical insurance system,the coverage of medical insurance has been continuously expanded.However,with the rapid development of medical insurance industry,some hidden problems have gradually emerged,such as the increasingly aging problem,more and more fiercely fraudulent behaviors and so on.These problems seriously threaten the normal operation of the medical insurance funds.Therefore,how to effectively identity potential risk factors,timely make early warning and take risk prevention and control measures in a targeted way becomes the key to maintain the normal operation of medical insurance system.In this dissertation,we take the risk prevention and control of basic medical insurance funds for urban workers in X city as the research object.The main objective is to comprehensively analyze potential risk factors affecting the balance of medical insurance funds,study key algorithms of balance risk prediction and fraud detection,establish risk prevention and control platform.The main contributions of the dissertation are as follows:· Time series-based balance risk prediction.Focus on the prediction of fund’s income and expenditure,we propose a univariate-based income forecasting model and a multivariate-based expenditure forecasting model based on time series prediction methods.By multiple comparative experiments on real medical insurance data,the effectiveness of two methods is proved.· Fraud risk detection based on local outlier factor.Aimimg at the problem of abnormal medical expense detection in fraud risk detection,based on the local outlier factor,by introducing DBSCAN clustering method and redefining the way to calculate the local outlier factor,two abnormal expense detection methods of FDC and FDP are proposed.And we analyze the fraud detection rate and time cost to verify the effectiveness of two methods.In addition,the introduction of rule-based abnormal frequency detection method makes it possible to supervise fraud detection in the whole process.· Research and implementation of risk prevention and control platform for medical insurance funds.Based on data warehouse and data mining technologies,introducing Spark distributed computing framework,MVC design pattern and Java EE development framework,we design and implement the medical insurance fund risk prevention and control platform.This platform mainly includes query and statistics,analysis and mining,earning and monitoring,management and maintenance,which can assist in decision-making and management for medical insurance funds.