Medical Service Demand Forecasting Based On Multi-Channel Data Analysis

IE is an engineering discipline based on quantitative analysis that emphasizes systemic and scientific,analyzes problems and solves problems from a system perspective.This idea of quantitative analysis is widely used in decision analysis and resource scheduling problems in the medical field.In the field of medical forecasting,medical service demand forecasting is the premise of scheduling and resource allocation in the operation management of medical institutions.Accurate forecasting of medical service demand is beneficial for the reasonable healthcare resource.It is of great significance to the scientific management of medical institutions.This paper takes the research status of medical service demand forecasting method as the starting point,and refines the medical service demand forecasting problem into medical independent demand forecasting and related demand forecasting.This paper separately studies different forecasting methods applied to these two kinds of problems.For independent demand forecasting.In this paper,the advantages and disadvantages of existing time series models are analyzed,and a hybrid forecasting model based on the autoregressive integrated moving average(ARIMA)model and the self-adaptive filtering method was presented to effectively improve the accuracy of time series prediction.It emphasizes the advantage of ARIMA model for feature identification and parameter estimation of time series data,and introduces the “weight” adjustment of the self-adaptive filtering method to adjust the parameters of ARIMA,which can improve forecast accuracy.In this paper,totally 13 cases of 4 typical time series data in different scales are selected from the time series data library(TSDL)as the test cases to compare the forecasting result by,the hybrid model and the ARIMA.Results shows that the prediction accuracy of the hybrid model is increased about 80%-99% than the ARIMA model for short-term forecasting,and the prediction tendency is closer to actual situation.As for multiple-stepahead forecasting,the rate of change of forecast PE of ARIMA model shows a linear increasing trend while than that of the hybrid model remains nearly unchanged,which indicates the effectiveness of long term forecasting of hybrid model.Subsequently,the number of daily prenatal examinations and the number of B-overexaminations in a maternal and child health care center(MCHCC)in Ningbo,China from January2017 to March 2018 were collected,and the hybrid model and ARIMA model were used forprediction.The prediction results show that the MAPE prediction results of the ARIMA model are18.53% and 27.69%,respectively,and the hybrid models are 2.79% and 1.25%,respectively.The results show that the hybrid prediction model is simple to implement and the prediction accuracy is better than the ARIMA model.At the same time,the applicability of the hybrid forecasting model in the visits forecast of MCHCC was studied.It was found that when the historical data amount was10-13 months,the forecasting accuracy was the highest,and when the historical data amount remained unchanged,the forecasting effect of daily inspection volume in October was the worst.For the related demand forecasting problem,the influence of weather and holiday factors on the number of outpatients was considered and as a correlation factor,a multi-channel data prediction model was established.This paper studies the ensemble decision tree forecasting algorithm,GBDT and random forest(RF),which is currently used in the popular machine learning method,and applied to the prediction of the number of daily daily prenatal examinations in MCHCC,and the modeling steps are elaborated.The forecasting results show that MAPE of GBDT and RF prediction results are 12% and 15%,respectively,which meet the application requirements.The model can be used in outpatient volume forecast.At the same time,the forecast effect of ARIMA,decision tree and hybrid forecasting model is evaluated from the PE,MAPE and standard deviation of PE of forecasting results.The evaluation results show that the hybrid forecasting model proposed in this paper outperforms the ensemble decision tree algorithm,and the traditional ARIMA model has the worst forecasting effect.