中国输血杂志2025，Vol.38Issue(2)：257-262,6.DOI:10.13303/j.cjbt.issn.1004-549x.2025.02.016
三种时间序列模型预测红细胞类血液需求趋势效果对比
Comparison of the effects of three time series models in predicting the trend of erythrocyte blood demand
摘要
Abstract
Objective To analyse and predict the tendencies of using erythrocyte blood in Changsha based on the au-toregressive integrated moving average(ARIMA)model,long short-term memory(LSTM)and ARIMA-LSTM combination model,so as to provide reliable basis for designing a feasible and effective blood inventory management strategy.Methods The data of erythrocyte usage from hospitals in Changsha between January 2012 and December 2023 were collected,and ARIMA model,LSTM model and ARIMA-LSTM combination model were established.The actual erythrocyte consumption from January to May 2024 were used to assess and verify the prediction effect of the models.The extrapolation prediction ac-curacy of the models were tested using two evaluation indicators:mean absolute percentage error(MAPE)and root mean square error(RMSE),and then the prediction performance of the model was compared.Results The RMSE of LSTM mod-el,optimal model ARIMA(1,1,1)(1,1,1)12 and ARIMA-LSTM combination model were respectively 5 206.66,3 096.43 and 2 745.75,and the MAPE were 18.78%,11.54%and 9.76%respectively,which indicated that the ARIMA-LSTM combination model was more accurate than the ARIMA model and LSTM model,and the prediction results was basi-cally consistent with the actual situation.Conclusion The ARIMA-LSTM model can better predict the clinical erythrocyte consumption in Changsha in the short term.关键词
ARIMA模型/LSTM模型/ARIMA-LSTM模型/需求/预测/红细胞Key words
ARIMA model/LSTM model/ARIMA-LSTM model/demand/prediction/erythrocyte分类
临床医学引用本文复制引用
邱亚娟,张建平,罗佳,李培林,罗孟卓,李琼英,刘格,雷情,廖凯..三种时间序列模型预测红细胞类血液需求趋势效果对比[J].中国输血杂志,2025,38(2):257-262,6.基金项目
湖南省卫生健康委 2024 年度卫生科研课题(W20243276) （W20243276）
