The present White Paper chooses to approach the advancement of Big Data and AI in healthcare by focusing on a major trend in digital health: Federated Learning (FL) and Federated Analysis (FA). This approach proposes to harness the full potential of health data by enabling the secure exploitation of multiple data sources without having to pool data in a single site (AbdulRhaman, 2020). FL/FA can be presented as a response to present legal, ethical, and technical challenges that limit data sharing across institutions and jurisdictions and thereby reduce the capacity to conduct collaborative data-driven research at a national and international scale (Kairouz et al., 2021).
While FL/FA presents genuine opportunities for the enhancement of Big Data and AI for research and innovation, this approach also raises several questions regarding privacy protection, data reliability, and resource utilization, among others. These are the specific issues investigated in this document. While exploring the potential and challenges of FL/FA for collaborative research in digital health, this white paper also describes robust platforms and technologies showing how FL/FA can be made possible in today’s healthcare systems.
The collaboration between the RLS-Digital Health members has shed light on projects that stand out as powerful examples of the promises of FL/FA for data-driven research and innovation in healthcare. Based on these inspiring initiatives, this white paper presents key conditions that could drive the establishment of successful infrastructures able to connect and analyze high quality and real-time data sources, while ensuring that the best standards for data protection and normalization are in place. These conditions could help us define and build a model for data-driven collaborative research at the national and international scales that could benefit researchers, innovators, decision-makers, and patients.
Read the full paper here: https://zenodo.org/records/10366653
