摘要/Abstract
摘要: 目的·利用差分整合移动平均自回归模型(autoregressive integrated moving average model,简称ARIMA模型)建立临床红细胞需求预测模型,为制定血液采集及献血者招募计划提供科学依据。方法·收集2006年1月—2016年6月重庆市血液中心的临床红细胞出库数据,利用SPSS软件建立ABO血型的ARIMA模型。运用建立的模型预测2016年7月—12月ABO各血型红细胞的临床需求量,并与实际值比较以验证预测效果。结果·成功建立ABO血型各自的最优模型,A型为ARIMA(3,1,0)(1,1,0)12、B型为ARIMA(3,1,0)(0,1,1)12、O型为ARIMA(3,1,0)(1,0,1)12及AB型为ARIMA(3,1,0)(0,1,1)12。各模型的Ljung-Box Q统计量均无统计学意义,残差序列均为白噪声,拟合效果良好;且各模型于2006年1月—2016年6月的拟合值与实际值间动态趋势大致相同,2016年7月—12月的临床需求量实际值均在预测值的95%CI内,平均相对误差均在10%以内。结论·ARIMA模型预测精度较高,能较好地拟合临床红细胞需求量的变化趋势,适用于临床红细胞各血型需求量的短期预测。
关键词: 时间序列分析, 临床红细胞, 需求预测, ABO血型
Abstract:
Objective · To establish the clinical usage demand prediction model of red blood cells (RBCs) based on the autoregressive integrated moving average (ARIMA) model, and provide scientific basis for plans of blood collection and blood donor recruitment. Methods · Clinical RBCs outflow data of Chongqing Blood Center from January, 2006 to June, 2016 were collected, and SPSS software was used to establish ARIMA model of ABO blood groups. The clinical usage demand of ABO blood groups of RBCs from July to December, 2016 were predicted by using the established model and the prediction effect was verified by comparing with the actual value. Results · The optimal models of ABO blood groups were established successfully, i.e. blood group A was ARIMA(3,1,0)(1,1,0)12, blood group B was ARIMA(3,1,0)(0,1,1)12, blood group O was ARIMA(3,1,0)(1,0,1)12, and blood group AB was ARIMA(3,1,0)(0,1,1)12. There was no statistical significance of Ljung-Box Q statistics in each model. The residual sequences were white noise with good fitting effects. The dynamic trend of the fitted value and the actual value from January, 2006 to June, 2016 of each ARIMA model was approximately the same, the actual values of clinical usage demand from July to December, 2016 were within the 95%CI of predicted values, and the average relative errors were less than 10%. Conclusion · The ARIMA model has a high prediction accuracy, which can fit the trend of clinical usage demand of RBCs well, and is suitable for short-term prediction of clinical usage demand of ABO blood groups of RBCs.
Key words: time series analysis, clinical red blood cell, demand prediction, ABO blood group
PDF全文下载地址:
点我下载PDF