邯郸地区临床红细胞用量数学模型的建立与预测研究  

The Establishment of Mathematical Model and Prediction Research on Clinical RBC Dosage in Handan Region

作　　者：李俊霞[1] 陈慧[1] 李源[1] 王洪[1] 孙国栋[1] 

机构地区：[1]河北省邯郸市中心血站,056001

出　　处：《临床输血与检验》2016年第3期212-214,共3页Journal of Clinical Transfusion and Laboratory Medicine

基　　金：河北省卫生和计划生育委员会医学科学研究重点课题计划(No.ZD20140390)资助

摘　　要：目的分析邯郸地区临床红细胞类制品使用规律,建立最优的数学模型并进行预测,以指导血液机构的相关业务工作。方法对邯郸市2002年1月~2013年12月每月向本地区临床供应红细胞类制品量,经Epidata3.0双录入数据,导入IBM SPSS 21,对临床用血量采用曲线回归和专家建模器分别建立数学模型,并选出最优模型,再利用该模型对临床用血量进行预测,并验证模型效果。结果曲线回归方程中R^2最高的曲线为三次多项式函数,R2=0.947,P<0.05,拟合方程为Y=2 413.906+83.189X-0.602X^2+0.004X^3。专家建模器给出ARIMA(0,1,1)(0,1,1)模型,残差的白噪声检验结果显示P>0.05,为白噪声序列,模型均提取了原序列中所有数据信息,模型诊断得以通过。用两种模型分别预测2014年1~6月临床用血量,ARIMA(0,1,1)(0,1,1)模型预测值的相对误差均在5%以内;三次方程模型预测偏差较大,最高达14.68%。ARIMA模型优于三次方程模型。结论通过建立数学模型的方式,补充后续数据,血液机构能够科学地预测用血趋势,有效指导采供血工作。Objective To analysis the regular pattern of using RBC class dosage for clinical use in Handan area, trying to set up the optimal mathematical model and make predictions. Then provide guidance for blood agencies＇ work related business. Methods Creating a dataset of the amount of clinical supplying red blood cells from January 2002 to December 2013 in Handan City with the support of Epidata3.0 double-lines input data system and then import it to IBM SPSS Statistics 21. Setting up a mathematical model for clinical using blood volume with curvilinear regression and expert modelers from which the optimal one is selected. Finally, reusing the model to predict clinical blood volume, and verify it. Results The highest R2 of curve regression equation as the cubic polynomial function. R2=0.947, P〈0.05, fitting equation =2 413.906＋83.189X-0.602 X2＋0.004X3. Expert modelers advised the model of ARIMA （0,1,1） （0,1,1）. The five models ofresidual white noise test results have shown the P〉0.05, instructions residuals all was white noise sequence, all models were extracted from the original sequence data information and passed the model diagnosis. With the two models respectively predict clinical using blood volume from January to June,2014, the relative error of ARIMA （0,1,1） （0,1,1） model prediction is within 5%; the one of cubic equation model, predicting the deflection, went up to 14.68%. ARIMA model is superior to the cubic equation model. Conclusion Through the establishment of mathematical model, adding the follow-up data, blood agencies can scientifically predict the trend of using blood volume and reasonably guide the corresponding business work.

关 键 词：临床用血 红细胞类 数学模型 预测 

分 类 号：R331.141[医药卫生—人体生理学]