Application and parameter optimization of the moving epidemic method for influenza surveillance and early warning

Abstract

Objective To explore the feasibility and parameter setting of moving epidemic method (MEM) in the surveillance and early warning of influenza. Methods The MEM models were established by using the percentages of influenza-like illness (ILI%) and positive rates of influenza virus (PR) between the 20th and 40th weeks of each year in Jingzhou City from 2010 to 2017. The optimal values of parameter δ were screened and the results of data fitting by both MEM models were compared. The evaluated indicators included sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, Matthew correlation coefficient and Youden index. Results For the optimal MEM model based on the ILI% data, the parameter δ was 3.0, and the sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, Matthew correlation coefficient and Youden index were 10.76%, 100.00%, 100.00%, 0.44, n/a, 0.89, 0.02, 10.76%, respectively. For the optimal MEM model based on the PR data, the parameter δ was 2.8, and the sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, Matthew correlation coefficient and Youden index were 79.62%, 95.90%, 93.33%, 86.73%, 19.44, 0.21, 0.78 and 75.53%, respectively. Conclusions The PR data was more suitable than the ILI% data for the influenza surveillance model by MEM in Jingzhou city. The optimal parameter δ was 2.8. Key words: Moving epidemic method; Influenza; Surveillance; Early warning