中国出生缺陷预测模型及比较

China birth defects prediction models and comparison analysis

目的运用GM、SPSS、SAS、EXCEL等软件建立我国出生缺陷发生率的灰色预测模型、回归模型、ARIMA模型,比较模型的优缺点并选择最优模型预测未来几年出生缺陷的变化,为母婴保健工作提供科学的依据。方法查找《中国卫生统计年鉴》2002-2011年出生缺陷发生率,用GM软件拟合GM(1,1)模型,用SPSS拟合回归模型,用SAS拟合ARIMA模型。结果灰色模型为y(t)=5 961.7e^(0.0218(t-1))-5 849.58,残差平方和为119.83;回归模型为Y=0.091X×3-2.103X×2+17.4X+97.928,残差平方和64.43;该序列为白噪声序列无法建立ARIMA模型。结论回归模型的精确度更高。对未来几年的出生缺陷进行预测,我国的出生缺陷发生率总体呈上升趋势,相关部门应积极采取干预措施降低发生率。

Objective To establish grey forecasting model,regression model and ARIMA model using GM,SPSS,SAS,EXCEL and other softwares to predict the incidence rates of birth defects in China,compare the advantages and disadvantages of these models,select the optimal model to predict the change of birth defects in the next few years,provide scientific basis for maternal and infantile health care work.Methods Based on the incidence rates of birth defects reported by China＇s Health Statistics Yearbook 2002-2011,GM（ 1,1） model,regression model,and ARIMA model were matched by GM software,SPSS software,and SAS software,respectively. Results Grey model： y（ t） = 5 961. 7e（0. 0218（ t-1））- 5 849. 58,the residual sum of squares was 119. 83; regression model： Y = 0. 091 X × 3- 2. 103 X × 2 ＋ 17. 4X ＋97. 928,the residual sum of squares was 64. 43; this sequence was white noise sequence,ARIMA model couldn＇t be established. Conclusion The accuracy of regression model is higher. The model predicts that the overall incidence rate of birth defects in the next few years shows an upward trend,the relevant departments should actively take intervention measures to reduce the incidence rate.