Identifying key factors in predicting Chikungunya and Zika transmission in French Polynesia: a data-driven mathematical model

BackgroundChikungunya and Zika are both arboviruses transmitted through the Aedes mosquitoes, which are ectothermic, leading to seasonal outbreak patterns of virus infections in the human population. Mathematical models linked with mosquito trap data, human case data, or both, have proven to be powerful tools for understanding the transmission dynamics of arboviral diseases. However, while predictive models should consider a variety of features in the environment, vectors, and hosts, it is not clear which aspects are essential to assist with short-term forecasting. MethodologyWe consider four simple models with various assumptions, including mosquito dynamics, temperature impacts, or both, and apply each model to forecast the Chikungunya and Zika outbreaks of nine different regions in French Polynesia. We use standard statistical criteria to compare the accuracy of each model in predicting the magnitude of the outbreak to select the most appropriate model to use as an alert system for arbovirus infections. Moreover, by calibrating our "best model", we estimate biologically meaningful parameter values to explore the commonality and difference between Chikungunya and Zika epidemics. ConclusionsWe show that incorporating the mosquito population dynamics in the arbovirus transmission model is essential for accurate arbovirus case prediction. In addition, such enhancement in the accuracy of prediction is more obvious for the Chikungunya data than the Zika data, suggesting that mosquito dynamics play a more important role in Chikungunya transmission than Zika transmission. In contrast, incorporating the effects of temperature may not be necessary for past outbreaks in French Polynesia. With the well-calibrated model, we observe that the Chikungunya virus has similar but slightly higher transmissibility than the Zika virus in most regions. The best-fit parameters for the mosquito model suggest that Chikungunya has a relatively longer mosquito infectious period and a higher mosquito-to-human transmission rate. Further, our findings suggest that universal vector control plans will help prevent future Zika outbreaks. In contrast, targeted control plans focusing on specific mosquito species could benefit the prevention of Chikungunya outbreaks.