基于ARIMA乘积季节模型的红细胞临床用量预测

Prediction of clinical usage of red blood cells based on multiple seasonal ARIMA model

目的:构建红细胞临床用量的ARIMA乘积季节预测模型,为科学制定采供血计划提供参考。方法:利用重庆市血液中心2006年1月至2018年12月的红细胞临床月用量数据,构建红细胞临床月用量ARIMA乘积季节预测模型,并对其进行识别、检验与优化,选择最优模型;再以2019年1~6月的临床月用量数据评估模型的预测效果。结果:获得的最优预测模型ARIMA(3,1,0)(0,1,1)12可较好地拟合预测红细胞临床月用量,经检验,其残差序列的ACF与PACF基本都落入95%CI内,Ljung-Box Q=14.749,P=0.396,提示模型残差呈白噪声。2019年1~6月红细胞临床实际用量基本落入预测结果95%CI内,且预测结果与同期红细胞临床实际用量基本吻合,平均相对误差为6.15%。结论:模型ARIMA(3,1,0)(0,1,1)12可用于预测红细胞临床用量的变化规律,为无偿献血招募、采供血及血液动态库存管理提供一定的依据。

Aim:To construct a multiple seasonal ARIMA model of clinical red blood cells(RBC)usage,so as to provide guides and references for scientifically formulating collection and supply plans.Methods:The data of monthly clinical RBC usage from January 2006 to December 2018 were used to establish a multiple seasonal ARIMA model of clinical RBC usage,and the optimal model was selected by identifying,testing and optimizing;then,the data from January to June 2019 were used to evaluate the prediction effect of the model.Results:The obtained optimal prediction model ARIMA(3,1,0)(0,1,1)12 could better fit and predict the monthly clinical RBC usage,the auto-correlation function and partial auto-correlation function of residual of the model were almost fall into the 95%CI,and Ljung-Box Q was 14.749,P=0.396,indicating that the model residual was white noise.Actual clinical usage of red blood cells from January to June 2019 largely fell within the 95%CI of the forecast results,and the forecast results largely matched the actual clinical usage of red blood cells over the same period,with an average relative error of 6.15%.Conclusion:ARIMA(3,1,0)(0,1,1)12 can be used to predict the variation of the clinical usage of RBC,and provide a certain scientific basis for voluntary blood donation recruitment,blood collection and supply,and dynamic blood inventory management.