Application of Artificial Intelligence in Radiology services to strengthen diagnostic access in underserved communities: A scoping review

Abstract

More than 65 million Americans live in Health Professional Shortage Areas where access to diagnostic imaging is severely limited or absent, contributing to an estimated 795,000 deaths or permanent disabilities annually from diagnostic errors — the single largest source of preventable serious harm in U.S. healthcare. Artificial intelligence (AI)-assisted radiology tools, including FDA-cleared detection algorithms, portable imaging platforms, and tele-radiology systems, offer a scalable pathway to address this crisis. Yet no comprehensive synthesis exists examining how these tools have been implemented in low-resource settings in ways applicable to U.S. community health infrastructure. This scoping review, conducted following PRISMA-ScR guidelines, systematically maps and synthesizes published evidence on AI-assisted diagnostic imaging deployment across low-resource and underserved health systems. Searches were performed across PubMed/MEDLINE, Embase, Cochrane Library, CINAHL, and grey literature repositories including WHO, CDC, HRSA, and NCI, covering English-language studies published between 2013 and 2025. Twenty-six studies across 14 countries were included, spanning breast cancer, lung disease, tuberculosis, cervical cancer, and cardiovascular conditions. Three dominant implementation models emerged: task-sharing frameworks in which non-specialist providers perform AI-assisted image acquisition; hub-and-spoke tele-radiology networks linking community acquisition sites to remote specialist interpretation; and mobile screening deployments with on-site AI triage. AI-assisted tools demonstrated diagnostic accuracy comparable to specialist review across multiple conditions, with reported reductions in time-to-diagnosis of 30–60%. Key barriers included infrastructure limitations, regulatory complexity, and reimbursement gaps; facilitators included government partnership, community health worker integration, and locally validated algorithms. Findings provide an evidence base for adapting AI-assisted radiology frameworks to U.S. Federally Qualified Health Centers, rural hospitals, and safety-net institutions, with direct implications for federal health equity priorities including the Cancer Moonshot Initiative and Healthy People 2030. Keywords: PRISMA-ScR, Federally Qualified Health Centers (FQHCs), BRIDGE Project, Artificial Intelligence.

Ähnliche Arbeiten