Integrating obstructive sleep apnea risk assessment with AI-enhanced lung cancer detection from CT imaging: a narrative review

Abstract

Abstract Background The convergence of artificial intelligence (AI) in medical imaging, accumulating evidence linking obstructive sleep apnea (OSA) to lung cancer, and the paradigm of opportunistic computed tomography (CT) screening raises the possibility of transforming single-indication imaging into a multi-disease assessment platform. Main body In this narrative review we examine, and critically appraise, the case for combining AI-enhanced lung cancer detection with automated OSA risk assessment from a single low-dose CT (LDCT) acquisition. We synthesise epidemiological data on shared risk populations—including contradictory and null findings—pathobiological mechanisms connecting intermittent hypoxia to tumour biology, and advances in deep learning for pulmonary nodule detection and upper-airway analysis, and we weigh the concept against established, low-cost OSA screening alternatives. We present this integration explicitly as a conceptual framework that has not been built or validated, rather than as a clinically ready tool. Conclusion Although the shared-risk rationale is biologically plausible, the epidemiological association remains debated, the anatomical and technical feasibility of deriving OSA risk from a chest LDCT is currently unproven, and prospective validation against existing screening tools is required before any clinical implementation. Clinical trial registration Not applicable.