Research And Implementation Of Medical Insurance Big Data Fund Forecasting Model

With the rapid growth of hospital expenses,health insurance funds in many areas of the country have been spent more than income.In order to improve this situation and ensure the safety and sustainable development of the health insurance fund.The Ministry of Human Resources and Social Security introduce the "Opinions on Controlling the Total Payment of Basic Health Insurance."According to The Global Budget Control policy,health insurance institutions set quotas for designated hospitals to control the unreasonable growth of their medical expenses.Therefore,the hospitals,as a major consumer of the medical insurance fund,the health insurance institution needs to consider the fund income and hospital spending capacity to allocate available funds.However,the existing way in which health insurance institution allocates health insurance funds lacks a scientific basis.Allocation is based on the hospital fixed rate of growth and this is likely to result in unreasonable distribution and waste of resources.From a statistical perspective,hospital cost prediction can be seen as a multivariate t:ime series analysis problem.However,the addition of external factors may weaken the model's simulation of the time series' own trends.How to enhance the trend simulation of the target sequence in multivariate time series prediction is a problem that needs to be studied and solved.In the case of data uncertainty or system instability,point predictions are affected by their poor interpretation.Often people use the prediction interval and the confidence interval to describe the range of values of real values,and quantify this uncertainty to make up for such defects.Describe the value range of hospital expenses to more accurately set up the medical insurance fund allocation plan.However,when making medical treatment decisions,medical insurance business personnel should not only consider the consumption capacity of designated hospitals,but also consider the income of medical insurance,and allocate available funds to various hospitals in the form of "receiving expenditures".This article has conducted in-depth research on the above various issues:1.A hospital cost prediction model based on multivariate time series is proposed.Firstly,the features are extracted from the medical insurance database by conventional methods,and then the important features are selected by the grey correlation analysis method.Then a time series similarity fast search algorithm(SFS)is proposed,which can find the most suitable time window and match the historical similar sequence by iteratively.New features are extracted from similar sequences.Based on the basic feature set selected from the external influence factors,new features are added,and a richer feature set is used for modeling prediction.At the same time,SARIMAX and LSTM are used to test the time series of different feature sets and different time granularities respectively,2.A time series prediction interval regression model(PIBS)based on integrated learning(Stacking)is proposed.The uncertainty is quantified by predicting the interval of hospital expenses.The models such as linear regression,SVR and LSTM are mainly used as the primary learner and secondary learner of Stacking.First of all,this paper proposes a method for constructing seasonal historical intervals.This method combines the uncertainty in the sequence period and outside the period,and uses the confidence interval to construct the time series history interval.Then,this paper proposes an interval comprehensive evaluation criterion(ICWC),which adds the requirement of interval centralization based on the width coverage criterion(CWC).Finally,the comprehensive evaluation criteria is used as the optimization goal,and the hospital cost range is predicted by using the integrated learning method(Stacking)considering three kinds of feature sets and multiple regression algorithms.The experiment proves that the model proposed in this paper can accurately predict the hospital cost interval,and lays a foundation for the later model of medical insurance fund allocation.3.A model of optimal allocation of medical insurance funds is proposed.How to rationally allocate medical insurance funds is the main goal of the model.Firstly,based on the prediction results of the first two models,based on the range of hospital consumption capacity and hospital cost,the difference between the fixed hospital fees is used as the loss function.According to the situation that the medical insurance business personnel may have to consider,different constraints are set,and the fixed combination of the quota of each hospital and the actual hospital cost is the smallest.At the same time,according to the difference of treatment efficiency between hospitals,this paper optimizes the distribution of medical insurance and hospital resources by optimizing the existing patient distribution,and finds the relationship between the change of patient distribution and medical insurance expenditure.The experiment proves that the medical insurance fund optimization allocation model can optimize the distribution of medical insurance funds,and can further optimize the fund's expenditure by changing the distribution of patients.Through the above research,this paper achieves an accurate prediction of the total hospital medical care,and evaluates the predicted value through the prediction interval.On the basis of accurate prediction of hospital medical total,consider the medical insurance expenditure and other conditions to obtain the optimal allocation plan of medical insurance fund;at the same time,save the expenditure of medical insurance fund by optimizing patient distribution,and provide auxiliary decision support for medical insurance institutions to rationally allocate medical insurance funds.