论文部分内容阅读
目的利用自回归滑动平均混合(ARIMA)模型预测长沙市流感样病例(ILI)的发病趋势。方法收集长沙市2006年第1周-2013年第10周由流感监测哨点医院每日报告的流感样病例监测资料,进行时间序列分析并建立预测模型,使用前364周资料建立模型,后10周资料评估模型预测效果。结果流感样病例监测资料构建ARIMA(1,0,0)模型,回归系数差异有统计学意义(P<0.05)。白噪声残差分析显示序列自相关函数的Box-Ljung统计量最小值为20.155(P>0.05),残差为随机性误差。1-364周资料所建立模型ARIMA(1,0,0)预测效果良好,实际值均在预测值的95%可信区间(95%CI)内,符合率达100%。2013第11-16周长沙市ILI%预测值分别为2.28%(95%CI:0.00%~6.21%)、2.31%(95%CI:0.00%~6.26%)、2.33%(95%CI:0.00%~6.30%)、2.35%(95%CI:0.00%~6.33%)、2.36%(95%CI:0.00%~6.35%)、2.38%(95%CI:0.00%~6.37%)。结论 ARIMA模型能较好模拟长沙市流感样病例的发病趋势。
Objective To predict the incidence of influenza-like illness cases (ILI) in Changsha using autoregressive moving average mixed (ARIMA) model. Methods The data of influenza-like cases monitored daily by influenza surveillance sentinel hospital from the first week of 2006 to the 10th week of 2013 in Changsha City were collected, analyzed by time series and the prediction model was established. The model was established 364 weeks before use, and the last 10 Weekly material evaluation model predicts the effect. Results The ARIMA (1, 0, 0) model was constructed based on surveillance data of influenza-like cases, and the regression coefficients were significantly different (P <0.05). The analysis of white noise residual showed that the minimum value of Box-Ljung statistic of sequence autocorrelation function was 20.155 (P> 0.05), and the residual error was random error. The model established by 1-364 weeks ARIMA (1,0,0) predicted good results, the actual values were within 95% confidence interval (95% CI) of the predicted value, with a compliance rate of 100%. The ILI% of Changsha in the 11th -16th weeks of 2013 was 2.28% (95% CI: 0.00% ~ 6.21%), 2.31% (95% CI: 0.00% ~ 6.26%), 2.33% % To 6.30%), 2.35% (95% CI: 0.00% to 6.33%), 2.36% (95% CI: 0.00% to 6.35%) and 2.38% (95% CI: 0.00% to 6.37%). Conclusion The ARIMA model can better simulate the incidence of influenza-like cases in Changsha.