Machine-learning triage in tertiary hospitals across Lagos: a 12-month cohort study
Abstract
We evaluated a machine-learning triage tool across three tertiary hospitals in Lagos over 12 months (n = 42,318 patients). The model, trained on Nigerian emergency-department records, achieved a specificity of 0.914 and reduced time-to-critical-care by 32% relative to standard triage. We discuss implementation lessons, ethical considerations, and generalization across West-African emergency contexts.
Keywords
Introduction
Emergency departments in Nigerian tertiary hospitals face acute pressure due to population growth and constrained resources. Rapid, accurate triage is central to patient outcomes…
A locally-trained ML triage tool cut time-to-critical-care by 32% while maintaining specificity above 0.91, with feasibility validated across 3 Lagos hospitals over 12 months.