Application Of Time Series Prediction In Drug Inventory Management Of Third-level Hospitals

Medical and health service is one of the major livelihood problems in the 21 st century.It is closely related to the health of hundreds people.The Ministry of Health explicitly called for "regular assessment of drug storage,more than 85% of the drug turnover rate of less than 10 to 15 days,regular assessment,analysis and improvement measures" in 2012.According to statistics,in China the average daily hospital turnover of drugs up to 34 Day,6 times in developed countries,the stock of drugs accounted for 30% to 60% of the total liquidity fund.It is not only hindering the further development of the hospital,but also contrary to the requirements of the Ministry of Health.Actually,the current HIS more manual use of high and low limit to automatically generate purchase orders.This way dependent on manager experience too much.Personal technical level and the status understanding degree of the hospital will directly affect the inventory turnover state.Based on the above background,this paper proposes an improved ARIMA model to realize the intelligent management of pharmaceuticals by combination of wavelet analysis and similarity measure.Based on the status quo and inventory requirements of medical drugs,three alternative management schemes are designed and analyzed empirically.The main results of this thesis are as follows:(1)Depth study the principle and the main content of time series prediction.In the context of real hospital operation,the nonlinear model is difficult to apply because it cost high and high cost of external factors.In this thesis,we choose ARIMA model based on wavelet analysis as the framework and combined pretreatment technology.Then we explore the main factors that can affect the prediction accuracy and propose corresponding improvement methods to reduce the error until the actual demand is reduced.The simulation results show that this model can effectively improve the prediction accuracy at the mutation nodes,and it has excellent fitting effect and accurate prediction effect in all types of datasets.(2)Based on the existing basic business of pharmacy management and in-depth understanding of the process of drug circulation and storage,this thesis proposed three inventory management program.The management plan take reduce the inventory turnover rate and to meet the cumulative rate of 85% of the standard as main propose and compare three plan from the turnover rate,turnover days,the cumulative number of days,out of stock rate,poor sales rate and other aspects of the effect.The empirical analysis shows that the first and second plan show good performance,if the criteria are in the strict acceptance stage,we can select the first one which shows more stable,else if in process of maintaining period which maintain a long period of time we can be considered plan B which shows higher efficiency.