Data Scientists Sylwana Kaźmierska and Joanna Piwko will present a case study of two projects Digica has developed and successfully deployed. We have selected medical projects for which data security and technical creativity were key.

  - The first introduced Federated Learning to a group of British hospitals. As an official supplier of AI services to the British government, we had a great opportunity to work with big data whilst preserving full patient confidentiality within the NHS. The stakes were high: models for tumor recognition that would cover all patient distribution.

  - The second project - medical tool recognition - seems straightforward until the complexity of the data is understood. We had to support over 100 surgical trays containing over 50 instruments each, many of which were similar. And... no actual dataset. To make things more complicated, we didn’t have physical trays in order that we could prepare a dataset. How did we achieve over 90% accuracy with synthetic data generation? How did we include support for new trays in our app within 24 hours of getting just 3D object files from our client?
Let's talk! Join us in Berlin for breakfast and networking :) 
Merantix digica breakfast