Artificial Intelligence Triage for Acute Stroke Thrombectomy Under Variable CT Perfusion Availability: A Distributionally Robust Evaluation of Healthcare Transfer Decisions

Acute ischaemic stroke triage with artificial intelligence (AI) must decide whether to transfer a patient to a comprehensive stroke centre when CT perfusion (CTP) is unavailable at the referring site, and the decision rule must remain defensible across the range of CTP availability that characterises real transfer networks. the present analysis formalises that decision as a distributionally robust optimisation over an ambiguity set of CTP-availability regimes, and characterises the operating characteristics of the before-after-control-impact (BACI) design that would be used to evaluate such a rule. No observational data are used or claimed: the analysis characterises the design under a declared variance structure (between-cluster SD = 6.0, residual SD = 4.5, intraclass correlation = 0.640), with power estimated across 2000 design cells. The design parameters were clusters of 8, 16, 24 and 32, and effect sizes of 3, 5 and 8 percentage points.52 clusters are required for 80% power at a 5-percentage-point effect, and that the minimum detectable effect at 80% power falls steeply as cluster count rises from 8 to 32. The contribution is a formal bridge between the robustness of an AI triage rule and the statistical power of the design used to evaluate it, and the paper argues that the two must be specified jointly rather than sequentially.