Achieving >90% Sensitivity in Forecasting Malaria Risk 12 Weeks in Advance in the Amazon

Abstract:

In the Americas, almost 90% of malaria is reported in the Amazon, where large-scale interventions reduced reported cases from 1.03 million in 2000 to 0.44 million in 2011. However, since 2011, malaria began to rebound and today, nearly all reductions achieved over the prior two decades have vanished. Between 2011 and 2017, Amazon-basin countries experienced a 167% increase in malaria cases, the largest percent increase compared to any other region in the world. Several factors contributed to this increase, including a strong El Nino Southern Oscillation (ENSO) in 2011-12 that produced favorable conditions for transmission, social unrest causing massive migration, rapid resource extraction leading to both vector habitat expansion and human exposure, and the end of the Project for Malaria Control in Andean Border Areas (PAMAFRO), which was supported by the Global Fund and provided comprehensive malaria control in Venezuela, Colombia, Ecuador, and Peru from 2006-2010. As governments in Latin America have taken greater responsibility for conducting vector-borne disease surveillance and control, there is an increasing need to use geospatial technologies and data to support real-time identification of potential malaria hotspots. With support from NASA HAQ, we have developed a multi-layered system that produces forecasts of malaria in real-time with high spatial resolution. Our system is capable of forecasting malaria outbreaks 12 weeks in advance with greater than 90% sensitivity. We use a land data assimilation system (LDAS), human population density model and weekly malaria surveillance to construct an ecologically-constrained regional model that produces outbreak forecasts, and a conditional, district-level Bayesian forecast model to improve spatial identification of risk. In addition, we have developed an Agent Based Model to evaluate different intervention scenarios, including insecticide-treated nets, indoor residual spraying, and prophylaxis. We are working with the Peruvian Centers for Disease Control and the Ecuadorian Ministry of Public Health to train and implement the system into their current infrastructure.