南通市流感样病例时间序列分析及发病趋势的预测研究

摘要:目的　探讨时间序列分析中的自回归求积移动平均模型（ARIMA）在南通市流感样病例（ILI）发病趋势预测中的应用。方法　收集“中国疾病预防控制系统”中2009年7月-2013年11月共238周流感样病例的发病监测数据，建立时间序列数据库，对每周流感样病例的发病人数进行ARIMA模型拟合，利用模型对2009年7月-2013年11月的周数据进行外部预测，并对2014年2月-2014年3月各周的流感样病例的发病情况进行前瞻性预测。结果　构建流感样病例周发病数的ARIMA（1，1，1）模型为（1-0.806B2）（1-B）lnXt＝（1-0.958B2）∝t，其中B代表后移算子，Xt代表ILI周发病数，εt为随机误差。外部预测的预测值和实际值基本相符，相对误差较小。前瞻性预测结果符合流感样病例的流行特征。结论　该模型能较好的模拟并预测南通市流感样病例的发病趋势。

Time series analysis and prediction model of incidence trend of influenza-like illness in Nantong

Abstract: Objective To forecast the incidence trend of influenza - like illness (ILI) in Nantong by autoregressive integrated moving average (ARIMA) model. Methods The time series database was created based on the data collected from the China information system for disease control and prevention from July 2009 to November 2013. The ARIMA model was constructed based on the number of monthly influenza-like illness cases for this period, and the model was used to predict the incidence cases from February 2014 to March 2014. Results The ARIMA model of the influenza-like illness cases was (1-0.806B2) (1-B) ln Xt= (1-0.958B2)∝t, in which B, t, Xt and εt denote back shift operator, time, number of weekly influenza-like illness cases and random error, respectively. The model fitted the data well and the prediction error is small. Conclusion The ARIMA model fits the trend of influenza-like illness cases well in Nantong City.