Data Federation in Healthcare for Artificial Intelligence Solutions

Abstract

Data federation offers a way to get data moving from multiple sources providing advantages in healthcare systems where medical data is often hard to reach because of regulations or the lack of reliable solutions that can integrate on top of protocols like FHIR, HL7, DICOM, among others. Given the increasing need for solutions that augment healthcare systems with artificial intelligence (AI), in fields like genomics, cancer treatment, and radiology, all of which will require solutions that can provide data at scale while being traceable, safe, and regulatory-compliant. This paper proposes an architectural solution that may provide the core capabilities to implement a data federation approach in a healthcare system to enable AI.

The Regulation of Artificial Intelligence in Digital Radiology in the Scientific Literature: A Narrative Review of Reviews

“…If we compare what emerges from the overview of reviews with the dedicated studies based on full scientific articles of 2021–2022 [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ], it emerges that there is a coincidence of issues, such as the problem of barriers and critical issues of the issues [ 10 ] in relation to the introduction of AI in the health domain, the need for a federative approach [ 11 ], the importance of ethical aspects [ 16 ], and the interest in local situations as in [ 13 , 16 ].…”

“…The need to overcome an analysis of regulatory realities with a patchy approach emerges strongly from this comparison with the studies based on full scientific articles [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ] but also from the survey reported above of some government websites, where the regulatory approach is described [ 43 , 44 ] also with reference to ethics [ 45 , 46 ].…”

“…In the last 2 years (2021–2022), there are 16 contributions including 8 reviews. The contributions based on full scientific articles [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ] dealt with various issues related to the regulations ranging from barriers for the radiologist [ 10 ], to the knowledge of this figure through surveys [ 13 ], to local experiences [ 12 , 13 , 14 , 15 , 16 ], to the implication of ethics [ 14 ], and the importance of federative activities [ 11 ]. Nair et al [ 10 ] highlighted that there are barriers that radiologists should be aware of prior to implementing Artificial Intelligence in daily practice.…”

“…Barriers include regulatory compliance, ethical issues, data privacy, cybersecurity, AI training bias, and safe integration of AI into routine practice. Castellanos et al [ 11 ] faced the importance of the data federation in this field. Data federation offers a way to get data moving from multiple sources, providing advantages in healthcare systems where medical data is often hard to reach because of regulations or the lack of reliable solutions that can integrate on top of protocols such as FHIR, HL7, and DICOM, among others.…”