预测医院门诊量的ARIMA模型构建及应用  

Construction and application of an ARIMA model for predicting the number of outpatient visits in general hospitals

作　　者：向前[1] 陈平雁[1] 

机构地区：[1]南方医科大学公共卫生与热带医学学院生物统计学系,广东广州510515

出　　处：《南方医科大学学报》2009年第5期1076-1078,共3页Journal of Southern Medical University

基　　金：国家自然科学基金(30571621);广东省科技计划(2005B33701005)

摘　　要：目的研究综合性医院月门诊量变化规律,预测其变化趋势,为医院管理决策提供依据。方法结合序列平稳性、长期趋势和季节效应,采用对数和差分变换,应用残差分析和最小二乘法估计,建立预测模型ARIMA(1,0,1)(0,1,1)12。结果拟合残差平方和为2.790,AIC=-178.126,SBC=-170.080,预测2008年门诊量相对误差为6.11%,小于指数平滑法(8.78%)。用该模型预测2009年门诊量为150.12万人次。结论医院门诊量存在季节变动和长期增长趋势,适合用ARIMA乘积模型进行拟合,但不同医院门诊量变化规律未必一致,要认真分析原序列的ACF图和PACF图,然后确定p、d、q参数。Objective To investigate the variation pattems of the number of outpatient visits in hospitals to provide references for more effective management of general hospitals. Methods The forecasting model of AR1MA （1,0,1）（0,1,1）12 was established using residual error analysis and least squares method according to the sequence stability, long-term trend and seasonal effect after logarithm transformation and differencing. Results The residual sum of squares was 2.790, AIC=-178. 126, SBC=-IT0.080. The relative predictive error of the model for predicting the outpatient visits in a general hospital in the year 2008 was 6.11%, smaller than that of exponential smoothing （8.78%）. This model predicted a number of outpatient visits of 1 501 200 in this hospital in the year 2009. Conclusions The ARIMA model provides a means for predicting the number of total outpatient visits, its long-term tendency and seasonal variation. The parameters p,d,q in the ARIMA model may vary between different hospitals, and the ACF and PACF charts of the original sequences are helpful for determining these parameters.

关 键 词：门诊量 ARIMA模型 门诊量预测 

分 类 号：R197.3[医药卫生—卫生事业管理]