应用CUSUM和EWMA模型探测甲型H1N1流感流行起始的预警研究

Study on early detection of the onset of the epidemic of pandemic H1N1 2009, using CUSUM and EWMA

目的探讨累计和（CUSUM）和指数加权移动平均（EWMA）模型探测甲型H1N1流感流行起始的功效。方法利用北京市2009年流感样病例监测数据，分别采用CUSUM和EWMA模型对数据进行分析，与流感病原学监测数据进行对比，分析了不同参数组合下2种模型探测甲型H1N1流感流行起始的效果。结果流感病原学监测显示2009年北京市甲型H1N1流感流行的起始时间为第41周，而CUSUM和EWMA模型均能在第42周之前发出预警信号；参数不同，CUSUM和EWMA模型的预警功效亦不同。结论CUSUM和EWMA模型参数值与其预警功效有关，CUSUM和EWMA模型可用于探测甲型H1N1流感流行的起始。

Objects To explore the efficiency of Cumulative Sum （CUSUM） and Exponentially Weighted Moving Average （EWMA） for detecting the onset of the epidemic of pandemic H1N1 2008. Methods Influenza like-illness （ILl） surveillance data was analyzed to detect the onset week of the epidemic of pandemic HIN1 2009, using the model of CUSUM and EWMA. The results were compared with the gold standard at different combination of parameter values. Results The 41st week was considered as onset of the epidemic of pandemic H1N1 2009 according to the gold standard based on the results from influenza virologic surveillance data. Both CUSUM and EWMA triggered the signal before the 42nd week. Differences in performances were observed as parameter values changed. Conclusions The parameter values of CUSUM and EWMA had a correlation to their performances. CUSUM and EWMA could be used for detecting the on- set of the epidemic of pandemic H1N1 2008.