4.2 CHIKRisk App: Global Mapping and Predicting Chikungunya Risk

Tuesday, 14 January 2020: 8:45 AM
153B (Boston Convention and Exhibition Center)
Radina Soebiyanto, GSFC, Greenbelt, MD; USRA, Greenbelt, MD; and A. Anyamba, R. Damoah, W. Thiaw, and K. Linthicum

Emerging and re-emerging diseases of global public health concern are recognized to be closely associated with variations in global climate. Recent chikungunya outbreaks in the Americas (2013-2016), Africa and Indian Ocean (2005-2006) and Asia have been associated with extreme departures in climate parameters including rainfall and temperature. Chikungunya in particular has illustrated the potential for global spread as demonstrated with epidemics in the Americas, Mediterranean Europe and by the limited local transmission in Florida and Texas. Under the umbrella of the Department of Defense (DoD) – Defense Threat Reduction Agency (DTRA) – Biosurveillance Ecosystem (BSVE) and NOAA International Research and Applications Project (IRAP) programs, we have developed a global chikungunya mapping and forecasting application system to map areas at risk for chikungunya concurrently and 1 to 3 months ahead. This risk mapping efforts take into consideration dynamic climate variables (rainfall and temperature) and static variables (population density and chikungunya vector presence). Chikungunya outbreak data compiled from publicly available sources (ProMED, Pan American Health Organization (PAHO), selected country health reports and Defense Health Agency weekly reports) are used to calibrate machine learning-based models. Random forest model was found to have highest accuracy in estimating areas at risk for chikungunya. Using this model, we are producing monthly risk maps based on climate observations and forecast risk maps based on NOAA’s North American Multi-Model Ensemble (NMME) temperature and rainfall forecasts. This effort is aimed at supporting DoD Force Health Protection (FHP) mission, regions of the US at risk (Texas and Florida) and international public health agencies (including World Health Organization, PAHO). This nascent effort illustrates how massive amounts of climate datasets combined with publicly available outbreak information using machine learning methods, can be brought to address an issue of public health concern. This effort also provides a template that can be employed in the immediate and near future to develop applications relevant to other vector-borne and ecologically coupled diseases.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner