详细信息
基于SARIMA模型的兰州市手足口病发病数预测
Prediction of onset number of hand-foot-mouth disease based on SARIMA model in Lanzhou
文献类型:期刊文献
中文题名:基于SARIMA模型的兰州市手足口病发病数预测
英文题名:Prediction of onset number of hand-foot-mouth disease based on SARIMA model in Lanzhou
作者:闫文博[1];蒋小娟[2];吴建军[1];殷晓春[1];周文瑜[1];沈棕林[1];王春庆[1];魏兴民[1]
第一作者:闫文博
机构:[1]甘肃中医药大学公共卫生学院,兰州730000;[2]甘肃省疾病预防控制中心,兰州730000
第一机构:甘肃中医药大学公共卫生学院
年份:2020
卷号:35
期号:2
起止页码:1
中文期刊名:疾病预防控制通报
外文期刊名:Bulletin of Disease Control & Prevention(China)
基金:甘肃省高校科研项目(2017A-058);甘肃中医药大学教学改革项目(YB-201615)。
语种:中文
中文关键词:手足口病;SARIMA模型;疾病预测
外文关键词:Hand-foot-mouth disease;SARIMA model;Disease prediction
摘要:目的建立甘肃省兰州市手足口病发病数的季节自回归滑动平均混合(SARIMA)模型,定量预测该地区手足口病的发病趋势,为疾病预警和风险评估提供依据。方法根据2009年1月-2017年12月兰州市手足口病报告发病数,采用SPSS 25.0建立SARIMA模型;利用建立的最优模型预测2018年1-12月兰州市手足口病发病数,并与实际发病数比较。结果兰州市手足口病以12个月为一个发病周期,每年有两个发病高峰期,分别是5-7月和9-11月,建立的最优模型为SARIMA(2,1,0)(0,1,1)12,各参数均有统计学意义(P<0.05),其平稳R2=0.361,R2=0.829,正态化BIC=9.618,该模型残差进行Ljung-Box检验结果为白噪声序列(q=16.408,P=0.355),与实际发病数比较,取绝对值后的相对误差为15.66%。结论建立的SARIMA(2,1,0)(0,1,1)12模型有效拟合了兰州市手足口病发病数时间序列,能够较好地预测兰州市手足口病发病趋势,对托幼机构、小学以及疾病防控单位提前采取应对措施具有一定的参考价值。
Objective To establish the seasonal autoregressive moving average(SARIMA)model of the onset number of hand-foot-mouth disease(HFMD)in Lanzhou city of Gansu province,so as to quantitatively forecast the epidemic trend of HFMD in this area,and provide basis for early warning and risk assessment of the disease.Methods Based on the reported number of HFMD in Lanzhou from January 2009 to December 2017,SARIMA model was established by SPSS 25.0.The number of HFMD cases was predicted with the optimal model,and compared with the actual onset number in Lanzhou from January to December,2018.Results In Lanzhou there was one epidemic cycle of 12 months of HFMD with 2 peaks of onset from May to July and September to November every year.The optimal model was SARIMA(2,1,0)(0,1,1)12,with parameters of statistical significances(P<0.05),stationary R2=0.361,R2=0.829,normalized BIC=9.618.The residual of the model was white noise sequence(q=16.408,P=0.355)tested with Ljung-Box method.Compared with the actual number of onset,the relative error was 15.66%after taking absolute value.Conclusions The model of SARIMA(2,1,0)(0,1,1)12 can effectively fits the time series of onset number of HFMD,fairly well to predict the epidemic trend of HFMD in Lanzhou city,and provide a certain reference for taking measures in advance in childhood cares,primary schools and institute of disease control and prevention.
参考文献:
正在载入数据...