Dengue Disease Vector Mapping via Environmental/Climatological/Sociological Factors

The United States (US) Government has established that a national/international biosurveillance capability is imperative for the safety and security of our nation. This imperative led to the recent establishment of a research and development effort sponsored by the Defense Threat Reduction Agency (DTRA) to develop an interoperable environment that will dramatically accelerate the nation's ability to detect, identify, and respond to a biological event of concern. This program, called the Biosurveillance (BSV) Ecosystem (BSVE), is bringing together data, algorithms, and analysts into a single modular and extensible system for meeting this challenge. The ultimate goal of BSVE is to improve incident outcomes for both natural and/or man-made (e.g. bioterrorist attack) disease threats. Vector-borne diseases, caused by pathogens that are transmitted by blood-feeding arthropods, are among the disease threats that the BSVE will need to address. For example, dengue, yellow fever, and chikungunya are caused by viruses transmitted primarily by the container-breeding mosquitoes Aedes aegypti (the principal virus vector in urban settings) and Aedes albopictus, both of which are widespread in the tropics and subtropics, including in the US. Among these diseases, dengue is already the most important arthropod-borne viral disease in the world, affecting ~390 million humans annually.

Despite their promise over purely empirical approaches, perhaps the greatest limitation in the current generation of weather-driven life cycle simulation models for Ae. aegypti (and other mosquito vectors) is their continued use of simplistic, statistically determined linkages between meteorological factors and mosquito bionomics to model population dynamics. With these approaches the water temperature in, water loss from, and sources of water for the immature mosquito stages (such as containers for Ae. aegypti or shallow surface water for anopheline malaria vectors) are modeled with statistical regression techniques that link them to the ambient weather conditions (e.g., air temperature, humidity, and cloud cover). These regression-based container water properties are then used to estimate mosquito populations and/or pathogen transmission risk within mosquito life cycle models. One significant hurdle for the regression-based methods is that the locally-determined relationships for container water dynamics rely on characterization of container types, sizes, and filling practices (which are linked to behavioral and socioeconomic factors), as well as local meteorological impacts (e.g., cloudiness). Because these factors have large spatial and temporal variability, it can be difficult to produce reliable results using the regression techniques outside of the region where the relationships were developed.

In this presentation, we will provide an overview of an approach to this problem that differs from the regression-based methods described above. Here we employ a physically-based modeling solution that involves a combination of:

1) data collected from a field survey conducted in August 2013 in Orizaba and Maltrata, Mexico; 2) the development of an image processing ("container detection") algorithm which searches commercially available high-resolution satellite imagery to locate and quantify container habitats; 3) the implementation of a recently-developed energy-balance model to simulate the amount and temperature of water in the containers where the immature stages of Ae. aegypti (and other arthropods) may occur; 4) the direct coupling of the energy-balance model with the container detection algorithm; 5) the linkage of the coupled container detection / energy balance model with publicly available models that simulate mosquito vector population dynamics and dengue virus transmission dynamics.

The combined system will simulate the abundance of the dengue virus vector mosquito Ae. aegypti and link this through models that simulate mosquito population dynamics and dengue virus transmission to provide a dynamically evolving map of the risk of contracting dengue. This physically-based approach takes into account the physical connections between Ae. aegypti, presence and prevalence of container habitats for the immature stages, and meteorological variables, as opposed to empirically-based approaches which rely entirely on statistical relationships that often "break" when they are implemented outside of the regions where they were developed. For this reason, we believe that the model framework developed through this effort may be extensible to a variety of locations globally, with different types or quantities of containers and climate conditions. This is currently an ongoing project and the presentation will cover an overall description of the project and preliminary results from the field survey and modeling components of the effort completed to date.