Grant: Project Grant

Mosquito-borne diseases are thriving despite decades of large-scale control efforts. Aedes mosquitoes are responsible for transmitting arboviruses such as chikungunya, dengue, and Zika. The geographic distribution of these viruses has dramatically expanded in the last few decades and Colombia is one of the countries in the Americas most affected by these arboviral epidemics. The overall aim of this study is to use a multi-disease perspective to identify disease patterns, drivers, and accurate disease forecasting methods that allow clinical and public health services to proactively plan and implement control and prevention measures. Disease, sociodemographic, and environmental data will be used in this study, which will be available at the national and municipal level in Colombia. We will begin by exploring the individual and joint disease dependencies by quantifying the correlation and variation of reported cases over time and space with Poisson-lognormal mixture models. The Bayesian multivariate models will be extended to forecast weekly disease burdens for a period of one year. Machine learning algorithms such as support vector machine regressions and random forest regressions will be developed to forecast disease burden. Semi-mechanistic dynamic models will also be developed for burden forecasts and these models will capture disease transmission processes. All forecasting results will be compared using various metrics of evaluation. The findings have the potential to greatly add to our understanding of the epidemiological profiles of chikungunya, dengue, and Zika, which is necessary for their successful control. The identification of at-risk areas and time periods can be used to inform targeted, integrated Aedes control programs and education campaigns. The forecasts, if accurate, will provide clinical and public health systems in Colombia with the necessary lead time for informed, proactive decisions.