Foodborne norovirus outbreak usually poses high risks in coastal areas in China. Owing to the influence of multiple climatic factors, it demonstrates typical seasonality and the hotspots gradually expanded northwards from 2008 to 2018. However, the complex mechanism of the onset of outbreaks makes accurate prediction difficult. Thus, it is in necessity to construct a predictive model for foodborne norovirus outbreaks in coastal areas based on environmental and geographical variables. A novel predictive nonlinear autoregressive model with exogenous inputs model was developed using 11 years of environmental and foodborne norovirus outbreak data collected from coastal areas in China. Five input variables (temperature, precipitation, elevation, latitude, and longitude) were screened through stepwise regression analysis. The predicted model developed in this study was able to reproduce 88.53% of outbreaks reported to the National Public Health Emergency Event Surveillance System (PHEESS) in the model development and 100% of outbreaks reported in the independent cross-validation since the system was first launched in China. In particular, foodborne norovirus outbreaks might occur when the probability is >0.6. The findings of this study suggest that foodborne norovirus outbreaks could be accurately predicted in coastal areas in China using the developed predictive model on a daily basis. The model output is most sensitive to temperature, followed by precipitation, and locations. The application of this predictive model is promising to improve local hygiene management levels, prevent foodborne norovirus outbreaks, and reduce the disease and economic costs in coastal areas in China.
Keywords: epidemiological characteristics; foodborne norovirus outbreak; nonlinear autoregressive model with exogenous inputs (NARX) model; prediction.