Savics, a social organization focused on improving healthcare delivery in developing regions, needed to optimize how a small team of doctors is dispatched across vast territories in African countries. In the context of tuberculosis control, this meant prioritizing areas with a high likelihood of outbreaks—typically locations with dense living conditions or poor infrastructure. Manual identification of such regions was not scalable, prompting the need for a systematic, data-driven approach.
To address this challenge, Savics partnered with Sagacify to develop AI models capable of identifying high-risk areas using satellite imagery. The collaboration involved setting up a labeling environment, curating relevant datasets, and integrating the final models into Mediscout+, Savics’ intervention planning platform.
Sagacify built a set of AI models based on Sentinel II satellite imagery to detect environmental and socioeconomic indicators linked to tuberculosis risk. One model classifies neighborhoods by poverty level, while another detects objects such as mines—both known to correlate with higher tuberculosis incidence. Labeled data was created by Savics with support from a custom annotation tool deployed by Sagacify. The resulting models were integrated directly into Mediscout+, enabling Savics to make targeted dispatch decisions based on geographical risk patterns.
Feeling lost ?
Let's discuss over a coffee; our AI experts have you covered!
