Spatiotemporal modeling of schistosomiasis in Ghana: linking remote sensing data to infectious disease
Wrable, Madeline R.
Gute, David M.
MetadataShow full item record
Citation (published version)Madeline R Wrable, Alexander Liss, Alexandra Kulinkina, Magaly Koch, Nana-Kwadwo Biritwum, Karen Kosinski, David M Gute, Elena Naumova. 2017. "SPATIOTEMPORAL MODELING OF SCHISTOSOMIASIS IN GHANA: LINKING REMOTE SENSING DATA TO INFECTIOUS DISEASE." AMERICAN JOURNAL OF TROPICAL MEDICINE AND HYGIENE. 65th Annual Meeting of the American-Society-of-Tropical-Medicine-and-Hygiene (ASTMH). Atlanta, GA, 2016-11-13 - 2016-11-16.
More than 90% of the worldwide schistosomiasis burden falls on sub-Saharan Africa. Control efforts are often based on infrequent, small-scale health surveys, which are expensive and logistically difficult to conduct. The use of satellite imagery to predictively model infectious disease transmission has great potential for public health applications. The transmission of schistosomiasis, a disease acquired from contact with contaminated surface water, requires specific environmental conditions to sustain freshwater snails. If a connection between schistosomiasis and remotely sensed environmental variables can be established, then cost effective and current disease risk predictions can be made available. Schistosomiasis transmission has unknown seasonality, and the disease is difficult to study due to a long lag between infection and clinical symptoms. To overcome these challenges, we employed a comprehensive 15-year time-series built from remote sensing feeds, which is the longest environmental dataset to be used in the application of remote sensing to schistosomiasis. The following environmental variables will be used in the model: accumulated precipitation, land surface temperature, vegetative growth indices, and climate zones created from a novel climate regionalization technique. This technique, improves upon the conventional Köppen-Geiger method, which has been the primary climate classification system in use the past 100 years. These predictor variables will be regressed against 8 years of national health data in Ghana, the largest health dataset of its kind to be used in this context, and acquired from freely available satellite imagery data. A benefit of remote sensing processing is that it only requires training and time in terms of resources. The results of a fixed effects model can be used to develop a decision support framework to design treatment schemes and direct scarce resources to areas with the highest risk of infection. This framework can be applied to diseases sensitive to climate or to locations where remote sensing would be better suited than health surveys.