自回归滑动平均混合模型在红细胞供应量预测中的应用

Autoregressive integrated moving average model and its application in prediction of red blood cell supply

目的探讨建立月悬浮红细胞供应量预测模型,为采供血提供参考。方法分别以浦东血站2007年1月-2014年6月的月红细胞供应总量、A+、B+、O+和AB+4种血型红细胞月供应量作为标本建立自回归滑动平均混合模型(ARIMA),用2014年7月-12月的实际值作为检验标本,用建立的最优模型预测2015年1-6月本站红细胞月供血量。结果经过序列平稳化、模型识别、建模和模型检验等步骤,月红细胞供应总量建立ARIMA(1,1,1)(1,1,1)12模型,调整决定系数(R2)=0.82,平均绝对百分比误差(MAPE)=5.98;A+型红细胞月供应量建立ARIMA(1,1,2)(1,1,1)12模型,调整R2=0.81,MAPE=8.42;B+型红细胞月供应量建立ARIMA(4,1,0)(1,1,1)12模型,调整R2=0.84,MAPE=7.23;O+型红细胞月供应量建立ARIMA(1,1,1)(0,1,1)12模型,调整R2=0.83,MAPE=6.63;AB+型红细胞月供应量建立ARIMA(1,1,0)(1,1,1)12模型,调整R2=0.80,MAPE=8.41。结论 ARIMA是一种短期预测精度较高模型,可用于红细胞供应量的预测。

Objective To establish an Autoregressive Integrated Moving Average model （ARIMA） based on monthly supply of red blood cell （RBC） in Pudong Blood Station, and to provide guidance for the blood station. Methods Total monthly supply of RBC （A ＋, B ＋, O ＋ and AB ＋ ） from January 2007 to June 2014 was used as samples to establish an ARIMA model respectively. The established models were used to forecast RBC supply in Pudong Blood Station from January to June 2015. Results Five ARIMA model were established respectively. The total monthly supply of RBC was established as ARIMA（ 1,1,1 ） （ 1,1,1 ） 12. Its adjusted R-squared value was 0.82. Its mean absolute percent error （MAPE） was 5.98. A ＋ RBC was established as ARIMA（ 1,1,2） （ 1,1,1 ） 12. Its adjusted R2 = 0. 82 and MAPE = 8.42. B ＋ RBC was estab- lished as ARIMA （4,1,0） （ 1,1,1 ） 12. Its adjusted R2 = 0.84 and MAPE = 7.23. O ＋ RBC was established as ARIMA （ 1,1, 2） （0,1,1） 12. Its adjusted R2 = 0. 83 and MAPE = 6. 63. AB ＋ RBC was established as ARIMA（1,1,0） （ 1, 1,1 ） 12. Its ad- justed R2 = 0. 80 and MAPE = 8. 41. Conclusion ARIMA is a short-term forecasting model which can be used in the predic- tion of the red blood cell supply.