Vector-borne disease transmission is highly variable over space and time, making it challenging to allocate resources for disease prevention, vector control and outbreak response. These patterns are often associated with variations in meteorological conditions that influence the reproduction, growth, and mortality of vector and host population along with the development and transmission of disease-causing pathogens. Monitoring environmental conditions with ground-based sensors and Earth-observing satellites can provide advance warning of disease outbreaks. However, predictive modeling of environmentally sensitive diseases is complicated by the complexities of transmission cycles and the effects of other factors related to human behavior and social environments.


Malaria remains one of the most significant public health challenges in many low- and middle-income countries, particularly in sub-Saharan Africa. To make surveillance a more effective and efficient core intervention, there is a need to shift from a reactive to a predictive approach by using models to forecast malaria risk. Our work on malaria early warning systems has involved developing and testing predictive models of weekly malaria cases in the epidemic-prone Amhara region of Ethiopia (Midekisa et al. 2012, Davis et al. 2019, Nekorchuk et al. 2021). We have also evaluated the accuracy of different sources of geospatial environmental data (Alemu and Wimberly 2020) and developed informatics tools to facilitate access to satellite Earth observations for malaria early warning (Liu et al. 2015; Wimberly et al. 2022). In collaboration with partners in the Ethiopian public health sector, we created the Epidemic Prognosis Incorporating Disease and Environmental Modeling for Integrated Assessment (EPIDEMIA) system for implementing malaria forecasting (Merkord et al. 2017) and conducted a scoping study to develop a roadmap for scaling malaria early warning systems to a national level in Ethiopia (Wimberly and Nekorchuk 2021). EPIDEMIA is implemented in the free and open-source R software for statistical computing and the code is freely available on GitHub. Visit the EPIDEMIA page for links to the software and documentation.


West Nile virus

West Nile virus is the most common mosquito-borne disease in the United States, causing more than 53,000 reported disease cases and 2,500 deaths since it was introduced in 1999. The Northern Great Plains have consistently been a hot spot of WNV incidence, with most cases concentrated in a few specific locations and occurring in a small number of outbreak years (Wimberly et al. 2013). Because of the relatively short disease transmission season, weather conditions in the spring and early summer influence the rate of virus amplification in mosquitoes and avian hosts and the subsequent risk of spillover into human populations (Wimberly et al. 2014). We developed a regional forecasting model that predicted WNV cases as a function of the rates of vegetation greenup and degree day accumulation during spring and summer (Chuang and Wimberly, 2012). This model was later refined to increase accuracy by incorporating both meteorological variables and mosquito infection rates as predictors (Davis et al. 2017, 2018). The approach was implemented as the Arbovirus Monitoring and Prediction (ArboMAP) system, which has been used to forecast WNV in South Dakota beginning in 2016. A retrospective validation of ArboMAP forecasts in South Dakota found that integrating environmental data with mosquito surveillance data increased the accuracy of WNV forecasts, and that accurate forecasts would be made early enough in the season to inform disease prevention and mosquito control activities prior to the peak of human transmission (Wimberly et al. 2022). Visit the ArboMAP page for links to the software and documentation.