Facebook IconFacebook IconFacebook IconFacebook Icon

Satellite imagery improves health care management

Sagacify Logo


Sagacify developed the Mediscout tool in collaboration with Savics. Savics, a social IT business, aim at providing adequate healthcare to everyone. Their product, Mediscout, uses satellite imaging to detect poor areas in Rwanda, Africa, which allows to identify precisely where their doctors should be sent in priority to maximize their impact.

The fight against tuberculosis

By analyzing satellite images analysis with machine learning models, we allowed Savics to quickly identify and locate people suffering from tuberculosis. Indeed, with tuberculosis, quick treatment is a key factor to keep the contamination rates low. And we know that the disease is more likely to break out in poor areas, which should then be reated first. The issue of the higher occurrence rate in poor neighborhoods is due to barriers (financial, geographical, gender-based…) that affect the accessibility of diagnostics and treatment. Consequently, we started developing the Mediscout tool which helps with planning, implementing and monitoring collaborative interventions in the fight against tuberculosis in Africa.


Convolutional Neural Networks

To distinguish poor neighborhoods from rich ones in satellite images, we chose to develop a model based on a convolutional neural network architecture (CNN) with transfer learning. The reasoning behind using a CNN is because these models easily detect patterns in images. The layers in a CNN are able to recognize certain features of images and then to combine them, creating more and more sophisticated patterns as we dig deeper into the layers of the network.


At the moment, the model can reach an accuracy of 94%. However, the system still has some trouble regarding unseen patterns (recognizing graveyard as poor area) and labeling issues (green area always recognized as poor) seem to surface as well. Above that, more classes other than poor, medium and rich should be added to provide more possible predictions. In the future, the model will be improved by feeding it with a larger pool of high-resolution images, executing post-processing on the model predictions while also extending the model to other African countries.

With the help of Mediscout, Savics now can efficiently dispatch its limited team of doctors to the areas that need immediate healthcare.

Related cases