RedraW
fedeRatED leaRning for humAn Wellbeing

The Project

The health care sector has been characterized by profound changes and large-scale adoption of heterogeneous digital solutions. An increasing number of citizens across Europe expect to access care quickly and easily choosing remote health smart devices to share with physicians own health data to generate information flows aimed at health care providers, to improve diagnosis, monitoring and treating processes of a growing number of conditions, mainly in frail patients. Public Health Administrations are performing increasingly relevant agreement to establish which technologies are most effective and which IT infrastructure can deliver safe, secure and interoperable access for collecting, processing, and distributing information flows from patients, but also healthy general population, to provide a sustainable predictive, preventive, personalized and participatory system. Individual European Countries has launched plans for the digitization of cross-border health services and the creation of a European health data space. A rational approach to the management of such a complex system cannot disregard the use of computational systems based on cloud resources and edge devices through heterogeneous and multi-level systems (cloud-edge continuum).

The Project

Within the REDRAW project, the development of research in this area is particularly relevant for at least two reasons:

• improving the technologies adopted for the monitoring, diagnosis and treatment management of specific health conditions;
  
  
• the need to develop approaches more respectful of the constraints of privacy, confidentiality, cybersecurity, accountability, and
  European digital sovereignty, which are still largely absent from the market.
  
  

REDRAW proposes the study and fine-tuning of dynamic cloud-edge deployment techniques, which exploits Federated Learning (FL)
models, in three real-world contexts, to improve the technological features of existing solutions, while respecting the strategic and
non-functional constraints that characterize the Italian and European scenarios.

The Case Studies

By the application of the FL paradigm into the cloud-edge continuum, REDRAW will enable the development of smart monitoring tools, which train locally a personalized model that learns how to evaluate information about the user’s well-being. The centralized
collection and elaboration of trained models will be exploited to learn how such conditions are evolving in the overall population without infringing the users’ privacy but allowing for deploying new models that consider global trends. The three case-studies investigate (1) maternal-fetal health care in pregnancy, conducted in collaboration with the Obstetrics and
Gynecology Clinic of the Bari Polyclinic, (2) treatment of neuromuscular disorders that characterize many genetic and degenerative diseases related to trauma or aging, in collaboration with the Neurological Clinic of the University of Pisa, and (3) monitoring of sleep disorders at IBB CNR of Naples.