综合性医院收容量预测的ARIMA模型构建研究

目的：研究综合性医院月收容量变化规律，监测医院收容量的异常变化，预测其变化趋势，为医院运营策略制定和资源调配提供依据。方法：通过对某医院1995--2005年月收容量数据分析，建立其监测和预测模型。结果：1995--2005年该医院住院病人月均收容2336±676．93人次，3、4、6、7、9、11和12月收容量超过10年月均水平，其余各月均低于月均水平。医院收容量的预测模型为ARIMA（0，1，1）（0，1，1）12，拟合残差平方和为2．810。以2005年月平均收容量预测值（2970±417，17）为目标值，2005年实际值超过目标值的26．12％。结论：该医院收容量存在季节效应和增长趋势，ARIMA模型不但可用于医院病人收容量的动态预测，还可用于医院病人收容量异常变化的监测和医院经营策略的评价，具有一定的实用价值。

Time Series Analysis by ARIMA Modeling to Forecast Amount of Inpatient

Objective： To help general hospital managers to make decision with the trend of amount of inpatient. Methods： The author performed time series analysis with an ARIMA （0, 1, 1 ） （0, 1, 1 ） 12 model which used monthly amount of inpatient data of a large general hospital from 1995 to 2004, and tested the model against the data in 2005. With the model, the managers can predict the trend and detect the abnormal changes in the amount of patient. Results： The average monthly amount of inpatient were 2 336± 676.93, and the amount in March, April, June, July, September, November and December exceeded the average. The residuals sum of square by fitting an ARIMA model to the 1995 -2004g amount of the inpatient was 2. 810. The study set the prediction of the average monthly amount of inpatient （2 970 ±417.17） in 2005 as a target value, and the result showed that the actual amount of inpatient exceeded the target value by 26.12 percent. The model could further be used to picture a control chart for early warning. Conclusion： There is an increasing trend in the hospital＇s amount of inpatient as well as the monthly pattern. The ARIMA model is suitable to forecast general hospital amount of inpatient and to find abnormal changing of it. Our approach potentially has a high practical value for hospital managers in decision making.