时间序列自回归移动平均模型在临床红细胞用量预测中的应用

目的验证自回归移动平均模型(ARIMA)预测临床红细胞用量的可行性,并为血站制定备血计划提供数据支持。方法选择东莞市2006年1月～2011年12月6年的每月临床红细胞用量作为时间序列模型的数据源。利用SPSS软件进行时间序列模型的构建,通过对2012年的前5个月临床红细胞实际用量进行模型检验。并据此对模型预测临床红细胞用量分析的可行性、建模步骤及准确性验证进行了探讨。结果 ARIMA模型计算出的预测值与实际值拟合较好,相对误差较小。1月份相对误差为-6.32%,2月份为13.28%,3月份为7.78%,4月份为3.73%,5月份为3.78%,平均相对误差为4.45%。结论可以应用时间序列自回归移动平均模型对未来的临床红细胞用量进行预测,为血站制定备血计划提供可靠的参考依据。

Time series autoregressive moving average model in the prediction of clinical use of red blood cells in Dongguan

Objective To study the feasibility of the autoregressive moving average model(ARIMA) in predicting the clinical usage of red blood cells(RBC),and provide information for inventory management of blood collection facilities.Methods The monthly use of RBCs in Dongguan city between January 2006 and December 2011 were taken to establish the model of time series data source.Using SPSS software to set up the time series model,the theoretical amount of blood use in Dongguan city for the first 5 months of 2012 was calculated and compared to the actual use of RBCs.Results The predicted values calculated by the ARIMA model and the actual values fit well,with only minor differences.The result of the mean of relative error was 4.45%,that is-6.32%(January),13.28%(February),7.78%(March),3.73%(April),3.78%(May).Conclusion Time series autoregressive moving average model may be use to predict the clinical RBC usage for blood collection facilities in the future.The model can provide a reliable reference for blood centers to make their blood supply plans.