Explainable Artificial Intelligence, Circular Resilience and Digital Interoperability in Healthcare Supply Chains: A Systematic Review and Integrative Research Agenda

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background: Healthcare supply chains determine the availability, safety and continuity of medicines, vaccines, devices and protective equipment. Explainable artificial intelligence, circular practices and digital interoperability may each strengthen these systems, but their evidence bases have developed separately. We reviewed how these three capabilities have been designed, implemented, evaluated and connected in healthcare supply chains. Methods: PubMed/MEDLINE, OpenAlex and Crossref were searched from inception to 1 August 2026 using four prespecified search families. Backward and forward citation searching supplemented database retrieval. Eligible records addressed an explained artificial-intelligence-enabled supply-chain decision, a circular practice explicitly linked to resilience, or demonstrated cross-system supply-chain data exchange. Two reviewers screened a calibrated high-sensitivity queue independently and a third adjudicated conflicts. Design-matched critical appraisal was applied. Heterogeneous evidence was synthesised narratively following reporting guidance for synthesis without meta-analysis, and cross-stream relationships were classified as directly tested, associated, theoretically argued or proposed. Results: The searches identified 2679 records, of which 438 were duplicates. Of 2241 unique records, 1702 were marked ineligible by prespecified automation-assisted rules and 539 underwent independent title and abstract screening. Eighty-five reports were assessed in full. The synthesis included 26 studies and three authoritative standards. Four studies addressed explainability, nine circular resilience and 14 interoperability, with one study contributing to two streams. Interoperability had the strongest field evidence and was associated with improved visibility and fewer vaccine stock-outs in one non-randomised implementation. Circular evidence mainly concerned controlled protective-equipment reuse during shortages. Explainability studies demonstrated technical explanations but did not test whether explanations improved human decisions. No study evaluated all three domains together. Conclusions: Interoperability is the most mature enabling layer. Circular practices can extend effective supply under controlled safety conditions, while explainable artificial intelligence remains technically promising but weakly evaluated with users. An integrated architecture should therefore be treated as a testable research proposition rather than an established intervention. Prospective multisite studies should measure availability, safety, resilience, environmental outcomes and explanation-supported decisions.