灰色模型GM(1,1)和残差自回归模型在中国孕产妇死亡率预测中的应用  

Application of grey model(1,1)and autoregressive error model in predicting maternal mortality rate in China

作　　者：张亚慧 张荣[1] 贾素红[1] 郭亮生[1] 张金婷 朱维培[1] ZHANG Ya-hui;ZHANG Rong;JIA Su-hong;GUO Liang-sheng;ZHANG Jin-ting;ZHU Wei-pei(Department of Obstetrics and Gynecology,the Second Affiliated Hospital of Soochow University,Suzhou,Jiangsu 215000,China)

机构地区：[1]苏州大学附属第二医院妇产科,江苏苏州215000

出　　处：《中国妇幼保健》2023年第10期1747-1752,共6页Maternal and Child Health Care of China

基　　金：2022年度省部共建放射医学与辐射防护国家重点实验室开放课题(GZK1202212);苏州大学附属第二医院优势学科托举项目(XKTJ-XK202006)。

摘　　要：目的采用灰色模型GM(1,1)和残差自回归模型分别对我国孕产妇死亡率(MMR)进行预测,并比较两者预测准确性,为MMR预测提供方法学指导。方法根据我国1991—2020年的MMR数据,应用灰色模型GM(1,1)和残差自回归模型,建立MMR预测模型并进行预测研究。结果GM(1,1)模型对MMR预测精度高于残差自回归模型,求得GM(1,1)方程为:x^((1))(k+1)=-1544.1427e^(-0.0511k)+1624.1427,拟合检验显示本模型拟合精度好(C=0.1868,P=1),能够较好地预测MMR的趋势,预测2022—2024年MMR分别为16.6/10万、15.8/10万和15.0/10万。结论对于1991—2021年MMR数据,GM(1,1)拟合效果优于残差自回归模型,预测表明2022—2024年我国MMR将继续呈下降趋势。Objective To predict the maternal mortality rate(MMR)in China based on the grey system model GM(1,1)and au⁃toregressive error model,and to compare the accuracy of the two models,and provide methodological guidance for MMR predic⁃tion.Methods According to the MMR in China from 1991 to 2020,the predicting model for MMR was established by using grey system theory GM(1,1)model and autoregressive error model to carry out predicting research.Results GM(1,1)model was more accurate on predicting the MMR than the autogressive error model.The grey model was x^((1))(k+1)=-1544.1427e^(-0.0511k)+1624.1427.The good⁃ness-of-fit test indicated that the precision was qualified(C=0.1868,P=1)and the grey model was capable in predicting the MMR in China.The predictive MMR was 16.6/105 in the 2022,15.8/105 in the 2023 and 15.0/105 in the 2024.Conclusion GM(1,1)model was more accurate on predicting the MMR from 1991 to 2021 than that the autoregressive error model.The predicting results indicates that the MMR will continue to decrease in China from 2022 to 2024.

关 键 词：孕产妇死亡率 灰色模型 GM(1 1) 残差自回归模型 预测 

分 类 号：R195.1[医药卫生—卫生统计学]