Data-driven Dispatching of Medical Doctors


Savics is a social organization that leverages technologies and on-the-ground knowledge to make an impact on global healthcare management. They are, among other things, active in the field of tuberculosis prevention and curing in African countries.

When planning their field operations, Savics is looking to optimize the time of a limited group of doctors, on a very large territory in order to visit in priority the places with a high probability of having a lot of cases. This is why they asked Sagacify to help them identify the best spots to visit thanks to earth observation, and AI.

Tuberculosis is known to be endemic in places with a high level of promiscuity, such as mines, poor neighborhoods, etc..  Savics asked Sagacify to automatically detect regions in Africa with these specificities (level of poverty, infrastructures elements, mines, etc) based on satellite images, using AI.


The developed models have been based on images taken by Sentinel II. Labelled dataset were created by the Savics teams, helped by a labellisation environment that was setup by Sagacify. 

The final models were integrated into Mediscout+, a satellite image solution developed by Savics to plan their interventions. 

Key elements

  • Neighborhood classification model on satellites images which determines the level of poverty of the neighborhood in african cities
  • Object detection model detects specific locations such as mines.   
  • Model deployed in production in Savics product


The resulting model is able to successfully identify the areas with a high likeliness of tuberculosis outbreaks. With the help of Mediscout+, Savics can now efficiently dispatch its limited team of doctors to the areas that need immediate healthcare.

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.

Related cases