Predicting the operations alert levels for dengue surveillance and control.

Operations alert level is a discrete measure that quantifies the severity of epidemic outbreak with respect to operational measures. The alert levels are ordered based on the amount of response operations required. In this paper, we develop multi-class classification models based on ordinal multinomial logistic regression for predicting the alert levels for dengue at twenty weeks in advance. The regression uses the dynamic lag non-linear models to account for the non-linearity of the dengue incidence, along with its lagged values. The performance of the models is tested for the dengue case count data of Singapore.