ARMA模型在医院入院量预测中的应用  

Application of ARMA Model in Prediction of Hospital Admission Volume

作　　者：罗利 陈柏旭 李佳玲 申文武[2] 朱婷 LUO Li;CHEN Boxu;LI Jialing;SHEN Wenwu;ZHU Ting(Business School,Sichuan University, Chengdu 610064;West China Hospital, Sichuan University, Chengdu 610041)

机构地区：[1]四川大学商学院,成都610064 [2]四川大学华西医院,成都610041

出　　处：《解放军医院管理杂志》2019年第7期605-607,612,共4页Hospital Administration Journal of Chinese People's Liberation Army

基　　金：四川省科技计划支撑项目(2016FZ0080)

摘　　要：目的研究大型三甲医院科室入院量变化规律,探究自回归滑动平均模型(Auto-Regressive and moving Aver-age Model,ARMA模型)预测入院量的可行性。方法利用ARMA模型拟合2015年4月-2016年4月肾脏内科入院量数据,并通过拟合所得模型对2016年5月该科室入院量进行预测并评价,建模与预测工具为Python3. 6。结果建立ARMA(0,3)模型,贝叶斯信息准则(Bayesian Information Criterion,BIC准则)为287. 533059,平均绝对误差百分比(Mean Absolute PercentageError, MAPE)为5. 7%,模型具有较高的预测精度。结论 ARMA模型能够以较高的精确度对科室入院人数进行短期预测。Objective To investigate the rule of admission volume in departments of large tertiary hospitals, and to explore the feasibility of predicting hospital admission volume by Auto-Regressive and Moving Average (ARMA) model. Methods The ARMA model was used to fit the data of admission volume of renal department from April 2015 to April 2016. The fit model was used to predict and evaluate the admission volume of the department in May 2016. The modeling and prediction tool was Python 3. 6. Results The ARMA (0, 3) model was established. The Bayesian Information Criterion (BIC criterion) was 287. 533059, and the mean absolute percentage error (MAPE) was 5. 7%. The model had high prediction accuracy. Conclusion The ARMA model enables short -term prediction of departmental admission volume with high accuracy.

关 键 词：ARMA模型 时间序列 入院人数 预测 

分 类 号：R197.32[医药卫生—卫生事业管理]