[1]于林凤),吴静),周锁兰),等.ARIMA季节模型在我国丙肝发病预测中的应用*[J].郑州大学学报(医学版),2014,(03):344.
 YU Linfeng,WU Jing,ZHOU Suolan,et al.Application of seasonal ARIMA model in forecasting incidence of hepatitis C in China[J].JOURNAL OF ZHENGZHOU UNIVERSITY(MEDICAL SCIENCES),2014,(03):344.
点击复制

ARIMA季节模型在我国丙肝发病预测中的应用*
分享到:

《郑州大学学报(医学版)》[ISSN:1671-6825/CN:41-1340/R]

卷:
期数:
2014年03期
页码:
344
栏目:
应用研究
出版日期:
2014-05-20

文章信息/Info

Title:
Application of seasonal ARIMA model in forecasting incidence of hepatitis C in China
作者:
于林凤1吴静2周锁兰1丁勇2)#
1)南京医科大学生物医学工程系 南京 2100292)南京医科大学数学与计算机教研室 南京 210029
Author(s):
YU Linfeng1 WU Jing2 ZHOU Suolan1 DING Yong2
1)Department of Biomedical Engineering, Nanjing Medical University, Nanjing 2100292)Department of Mathematics and Computer Sciences, Nanjing Medical University, Nanjing 210029
关键词:
ARIMA季节模型丙肝发病预测
Keywords:
seasonal ARIMA model hepatitis C incidence prediction
分类号:
R512.6
摘要:
摘要目的:应用ARIMA季节模型对我国丙肝发病进行预测。方法:利用2004年至2011年我国丙肝的月发病数建立ARIMA季节模型,对2012年丙肝的月发病数进行预测,并用实际数据评估模型的预测效果。同法对同期甲肝发病数据进行建模和预测。对丙肝和甲肝2004年至2011年的月发病数按年归一化处理后计算方差。比较甲肝和丙肝的预测效果。结果:成功建立ARIMA(1,1,1)(2,1,0)12季节模型,模型的表达式为:(1+0.222L)(1+0.820L12+0.694L24)(1-L)(1-L12)lnYt=(1+0.648L)εt,参数 AR(1)=-0.222(t=-2.392,P=0.020),SAR(12)=-0.820(t=-8.009,P<0.001),SAR(24)=-0.694(t=-6.124,P<0.001),MA(1)=-0.648(t=-5.889,P<0.001),残差序列是白噪声序列(P>0.05);模型拟合效果的R2为0.824,预测的平均相对误差为0.078。归一化后丙肝和甲肝发病数的平均方差分别为0.030和0.047,提示丙肝原始数据周期性动态变化较甲肝更趋一致。甲肝预测的平均相对误差为0.138,大于丙肝。结论:ARIMA(1,1,1)(2,l,0)12季节模型可用于预测我国丙肝的发病规律。样本数据的周期性动态变化趋势越一致,ARIMA季节模型的预测结果也越准确。
Abstract:
AbstractAim: To forecast the incidence of hepatitis C in China using seasonal ARIMA model.Methods: Seasonal ARIMA model was established based on the monthly reported cases data of hepatitis C in China from 2004 to 2011,and used to forecast the data of 2012.Actual data of 2012 were used to assess prediction effect. The model establishment and forecasting for hepatitis A were carried out using the same method. The variance of hepatitis A and hepatitis C incidence from 2004 to 2011 normalized according to the years was calculated. The predicted effect of hepatitis A and hepatitis C was compared.Results: The model of ARIMA(1,1,1)(2,1,0)12 was established successfully.The expression of the model was (1+0.222L)(1+0.820L12+0.694L24)(1-L)(1-L12)lnYt=(1+0.648L)εt,the parameters were as follows: AR(1)=-0.222(t=-2.392,P=0.020),SAR(12)=-0.820(t=-8.009,P<0.001),SAR(24)=-0.694(t=-6.124,P<0.001),MA(1)= -0.648(t=-5.889,P<0.001),residual error sequence was white noise sequence (P>0.05), the R2 of fitting was 0.824 and the averge error of prediction was 0.078. The averge variances of hepatitis C and hepatitis A normalized incidence were 0.030 and 0.047, suggesting that periodic dynamic change of hepatitis C data was more consistent. The averge relative error of prediction of hepatitis A was 0.138,higher than that of hepatitis C.Conclusion: ARIMA(1,1,1)(2,1,0)12 season model can be used to predict incidence of hepatitis C in China. Periodic dynamic change trend of sample data is more consistent, the ARIMA seasonal model predicted result is more accurate.

参考文献/References:

[1]中华预防医学会医院感染控制分会.中国丙型病毒性肝炎医院感染防控指南(2012年)[J].中国医学前沿杂志:电子版,2012,4(11):57 [2]Bowerman BL, O'Connell RT.Forecasting and time series:an applied approach[M].3rd ed.北京:机械工业出版社,2003:437 [3]方积乾,陆盈,张晋昕,等.现代医学统计学[M].北京:人民卫生出版社,2002. [4]Box GEP,Jenkins GM.Time series analysis: forecasting and control[M].HoldenDay,1976. [5]攸频,张晓峒.Eviews 6实用教程[M].北京:中国财政经济出版社,2008. [6]易丹辉.数据分析与EViews应用[M].北京:中国人民大学出版社,2008. [7]张彦琦,唐责立,王文昌,等.ARIMA模型及其在肺结核预测中的应用[J].现代预防医学,2008,35(9):1608 [8]胡建利,祖荣强,彭志行,等.江苏省戊型肝炎发病趋势的时间序列模型应用[J].南京医科大学学报:自然科学版,2011,31(12):1874 [9]胡建利,梁祁,吴莹,等.季节时间序列模型在菌痢发病预测中的应用[J].中国卫生统计,2012,29(1):34 [10]Muhammad HL,Nur Haizum Suhartono AR,Mohd TL,et al.Seasonal ARIMA for forecasting air pollution index:a case study[J].Am J Applied Sci,2012,9(4):570 [11]王丙刚,曲波,郭海强,等.传染病预测的数学模型研究[J].中国卫生统计,2007,24(5):536

备注/Memo

备注/Memo:
*江苏省大学生实践创新训练计划项目2012JSSPITP1033;南京医科大学基础医学院优势学科教师培养基金项目JX10131801099#通讯作者,男,1956年8月生,硕士,教授,研究方向:生物统计,Email:yding@njmu.edu.cn
更新日期/Last Update: 1900-01-01