Research question
Can geospatial algorithms be used by disease programs to help identify hotspots at community and Implementation Unit level?
Outcomes
· When conducting spatial prediction, the traditional method of selecting survey sites at random is inefficient when compared to spatially adaptive sampling. This is true whether the goal is to obtain as accurate estimates as possible or if the goal is to classify communities as requiring treatment on the basis of exceeding a prevalence threshold.
· As this and other studies now show, where decisions have to be made over regions such as districts or sub-districts, adaptive sampling can also improve these predictions in terms of root mean squared error or classification (above or below a prevalence threshold) accuracy.
· Results showed that making decisions at community level using predictions was more efficient, in terms of the number of individuals that would require treating for a given level of sensitivity, than treating at the sub-district or district level (as currently done). However, community-level administration may be too cumbersome in most areas.
Study sites
Regional Study Site, Cameroon
Notes
This project builds on work conducted in collaboration with the Task Force for Global Health, the Disease Prevention and Control Bureau, Department of Health, Philippines and the Ministry of Public Health and Sanitation, Haiti. While not a formal collaborator, we will also continue to work with Australian National University and will be supporting FHI 360’s Act to End NTDs | West Program to use the tool developed under this project.