Early warning of influenza epidemic based on CUSUM and EWMA models in Daxing District, Beijing
10.3969/j.issn.1006-2483.2026.01.003
- VernacularTitle:基于CUSUM和EWMA模型的北京市大兴区流行性感冒预警
- Author:
Hong LEI
1
;
Qiuling LI
1
;
Qi LIU
1
;
Meichen LIU
1
;
Enhuan DU
1
;
Jinfeng TANG
1
;
Zhiping LI
1
;
Yadi GAN
1
;
Lijie ZHANG
2
Author Information
1. Daxing District Center for Disease Control and Prevention, Beijing 100026, China
2. Chinese Field Epidemiology Training Program, Chinese Center for Disease Control and Prevention, Beijing 100050, China
- Publication Type:Journal Article
- Keywords:
CUSUM;
EWMA;
Influenza;
ILI;
Early warning
- From:
Journal of Public Health and Preventive Medicine
2026;37(1):13-17
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the effectiveness of the cumulative sum (CUSUM) and the exponentially weighted moving average (EWMA) for early warning of influenza epidemic using two datasets of reported influenza cases and influenza-like illness (ILI) cases. Methods Using the reported cases of influenza and ILI in Daxing District, Beijing, from week 23 of 2018 to week 22 of 2024 as data sets, the CUSUM and EWMA models were established, respectively. The positive rate of influenza etiology was used as the “gold standard”, and the Youden index was used as the evaluation index to compare the early warning effect of the two models under different data sets and different parameters. Results In CUSUM, the optimal Youden indices of the reported influenza cases set and the ILI cases set were 0.751 and 0.635, respectively. In EWMA, the optimal Youden indices of the reported influenza cases set and the ILI cases set were 0.544 and 0.464, respectively. The optimal EWMA and CUSUM models could both issue early warning signals in advance of the “gold standard”. Conclusion In the influenza epidemic early warning in Daxing District, Beijing, the CUSUM model established with the reported cases of influenza can achieve good early warning effects, but the model parameters need to be dynamically adjusted according to the local epidemic characteristics.
