Leveraging Federated Learning, Explainable Artificial Intelligence, and Governance Requirements for Privacy-Preserving Multisite Clinical Diagnostic Platforms: A Scoping Review Protocol

Abstract

This registration presents the protocol for a prospective scoping review of the technical, implementation and governance evidence needed to design, evaluate and govern privacy-preserving multisite clinical diagnostic artificial intelligence platforms. The review will examine federated-learning architectures and aggregation strategies, privacy-preserving mechanisms, explainability and interpretability methods, fairness auditing and bias mitigation, interoperability standards, validation and deployment practices, regulatory conformity, data governance, implementation barriers and facilitators, and equity considerations. The review will follow the Arksey and O’Malley framework, Joanna Briggs Institute guidance and PRISMA-ScR reporting recommendations. Literature published in English from 1 January 2017 onward will be searched in MEDLINE via PubMed, Embase, IEEE Xplore, the ACM Digital Library, Scopus, Web of Science and the Cochrane Library. Database searching will be supplemented by regulatory and standards documentation, preprint servers and backward and forward citation tracking. The review is expected to produce an evidence-gap map, a requirements-engineering maturity landscape, a regulatory-alignment summary, an analysis of implementation barriers and facilitators, and a prioritised, evidence-traceable framework of technical, governance and implementation requirements for federated and explainable diagnostic platforms intended for deployment across high-income and resource-variable healthcare settings. Formal searching, screening and data charting have not yet begun and will commence after prospective registration