回归滑动平均混合模型预测广州地区悬浮红细胞出库量的研究  

The analysis and prediction of the amounts of RBC issue in Guangzhou area based on ARIMA model

作　　者：罗伟峰[1] 林永桔[1] 郑晓晓[1] 钟惠珊 张旸 滕青[1] 聂咏梅[1] LUO Weifeng;LIN Yongju;ZHENG Xiaoxiao;ZHONG Huishan;ZHANG Yang;TENG Qing;NIE Yongmei(Guangzhou Blood Center,Guangzhou Key Laboratory of Medical,Guangzhou 510095,China)

机构地区：[1]广州血液中心广州市医学重点实验室,广东广州510095

出　　处：《中国输血杂志》2020年第5期472-475,共4页Chinese Journal of Blood Transfusion

基　　金：广州市医学重点学科建设项目(血液安全重点实验室),广州市卫生健康科技项目(20191A010048)。

摘　　要：目的建立准确预测血站悬浮红细胞出库量模型以便科学合理地安排采供血工作。方法收集2013—2017年广州血液中心血库每月出库供应给临床的悬浮红细胞悬液数(量)据,应用SPSS 25.0统计软件建立回归滑动平均混合模型(ARIMA),预测2018年1—9月红细胞悬液出库量,以届时实际的悬浮红细胞出库(临床供应)量评估ARIMA的预测效果。结果 2013—2017年广州血液中心的悬浮红细胞出库量持续增长,年均增长率3.91%,ARIMA(3,1,3)(1,1,0)12为相对最优模型,2018年1—9月悬浮红细胞出库量预测值为329 228 U,当时的实际值为309 837 U,第3季度误差相对较大(-11.43%),第1、2季度相对误差较小(-2.43%、-3.51%),平均相对误差为-5.89%。结论 ARIMA模型预测的血站短期(半年左右)悬浮红细胞出库(供应)量,可以作为血站悬浮红细胞库存量的参考阈值,为血站合理制定采供血计划提供科学工具。Objective To apply a model to accurately predict the amounts of RBC issue in order to scientifically and rationally arrange blood collection and supply for blood banks. Methods The amounts of RBCs issue for hospitals were drawn from the database in Guangzhou Blood Center during 2013 to 2017 by month. The ARIMA model was created by SPSS Statistics 25.0 software to predict the issuing amounts from January to September, 2018 and evaluate the effectiveness of the model based on the actual amount of RBC issue(clinical supply) at that time. Results The amounts of RBC issue from 2013 to 2017 in Guangzhou Blood Center had been increasing with an average growth rate of 3.91%. ARIMA(3,1,3)(1,1,0)12 was a relatively effective model. The predicted value of the amounts of RBC issue from January to September, 2018 was 329 228 U, and the actual value during the same time period was 309 837 U. The relative error between predicted values and actual values was relatively large(-11.43%) in the third quarter and marginal in the first and second quarter(-2.43% and-3.51%), with an average relative error of-5.89%. Conclusion The ARIMA model can be used to predict the amounts of RBC issue in a short term(about half a year), which can be used as a reference threshold for RBC inventory levels in blood banks,and provide a scientific tool to reasonably formulate a plan to collect and issue blood.

关 键 词：回归滑动平均混合模型 血站(库)血液出库量 悬浮红细胞 出库量预测 血库库存阈值 

分 类 号：R457.12[医药卫生—治疗学] O242.1[医药卫生—临床医学]