论文部分内容阅读
目的预测上海市流感样病例的发病趋势。方法应用自回归求和移动平均模型(autoregressive integrated moving average model,ARIMA),对上海市2004年1月至2006年4月期间由流感监测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114周资料建立模型,后9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)(1,1,0)26模型,非季节性和季节性自回归系数差异皆有统计学意义(P<0.001)。白噪声残差分析显示序列白相关函数的 Box-Ljung 统计量最小值为0.803(P>0.1),残差为随机性误差。1～114周资料所建立模型 lgy_t=0.879 lgY_(t-1)+0.418 lgY_(t-26)-0.367 lgY_(t-27)+0.582 lgY_(t-52)-0.512 lgY_(t-53)预测效果良好,实际值均在预测值的95%可信区间内,符合率达100%。结论 ARIMA 模型能较好模拟上海市流感样病例的发病趋势。
Objective To predict the incidence of influenza-like illness in Shanghai. Methods The autoregressive integrated moving average model (ARIMA) was used to analyze the data of influenza-like surveillance cases reported daily by the Influenza Surveillance Point Hospital from January 2004 to April 2006 in Shanghai The model was established by using the data of the first 114 weeks and the data of the last 9 weeks by the model to predict the effect. Results There was significant difference between ARIMA (1, 0, 0) (1, 1, 0) 26 model and non-seasonal and seasonal autoregressive coefficient in flu-like cases surveillance data (P <0.001). The analysis of white noise residual showed that the minimum value of the Box-Ljung statistic of the sequence white correlation function was 0.803 (P> 0.1), and the residual error was a random error. 1 to 114 weeks. The model lgy_t = 0.879 lgY_ (t-1) +0.418 lgY_ (t-26) -0.367 lgY_ (t-27) +0.582 lgY_ (t-52) -0.512 lgY_ (t-53) The effect is good, the actual values are within the 95% confidence interval of the predicted value, with a compliance rate of 100%. Conclusion ARIMA model can better simulate the incidence of flu-like illness in Shanghai.