中国媒介生物学及控制杂志 ›› 2012, Vol. 23 ›› Issue (4): 347-349.

• 调查研究 • 上一篇    下一篇

基于GM(1,1)-GRNN组合模型的肾综合征出血热发病率预测

吴昊澄, 王臻, 何凡, 孙继民, 曾蓓蓓, 鲁琴宝, 刘碧瑶, 赵艳荣   

  1. 浙江省疾病预防控制中心疾病控制应急办公室, 浙江杭州 310051
  • 收稿日期:2011-12-26 出版日期:2012-08-20 发布日期:2012-08-20

Prediction of the incidence of hemorrhagic fever with renal syndrome based on GM(1, 1)-GRNN model

WU Hao-cheng, WANG Zhen, HE Fan, SUN Ji-min, ZENG Bei-bei, LU Qin-bao, LIU Bi-yao, ZHAO Yan-rong   

  1. Zhejiang Center for Disease Control and Prevention, Hangzhou 310051, Zhejiang Province, China
  • Received:2011-12-26 Online:2012-08-20 Published:2012-08-20

摘要: 目的 预测浙江省肾综合征出血热(HFRS)发病趋势,为卫生工作的决策和防病治病提供科学依据。方法 根据浙江省2001-2009年HFRS发病率,建立灰色GM(1,1)与广义回归神经网络组合〔GM(1,1)-GRNN〕模型,与GM(1,1)模型对比,并进行预测。结果 2001-2010年HFRS年发病率组合模型拟合值与实际值的平均相对误差为2.47%,明显低于GM(1,1)模型的平均相对误差(11.19%),组合模型的预测精度相对单一GM(1,1)模型有较大的改善;组合模型预测2011年HFRS年发病率为0.8591/10万。结论 浙江省2010-2011年HFRS的发病率呈下降趋势,组合模型的预测精度高,可以提供更为准确的预测数据,从而为防控决策提供可靠依据。

关键词: 肾综合征出血热, GM(1,1)模型, 广义回归神经网络, 预测

Abstract: Objective To establish an effective model for monitoring and predicting the incidence trend of hemorrhagic fever with renal syndrome(HFRS)in Zhejiang province. Methods HFRS data from 2001 to 2009 in Zhejiang province were used to establish the GM(1,1)-Generalized regression neural network[GM(1,1)-GRNN]model. The GM(1,1)-GRNN model was used to predict the incidence trend of HFRS of the years 2010 to 2011. Results The predicted HFRS incidence with the GM(1,1)-GRNN model for the years 2001 to 2010 closely followed the observe value of the same years, with a relative error of 2.47%, more precise than that of GM(1,1) (11.19%). The forecast values for 2011 was 0.8591 per 105 population. Conclusion The GM(1,1)-GRNN model is highly precise in the prediction of the incidence of HFRS. It provides reliable accurate predicative data, contributing to decision-making regarding the control and prevention of the disease.

Key words: Hemorrhagic fever with renal syndrome, GM(1,1) model, Generalized regression neural network, Prediction

中图分类号: