Artificial Intelligence Triage for Ebola Virus Disease Under Variable Community Case-Definition Criteria: A Calibration-Transfer Evaluation of Isolation Unit Referral Timing

Community case-definition criteria for Ebola virus disease (EVD) vary across districts, and an artificial intelligence triage model calibrated under one set of criteria may transfer poorly to another, shifting the timing of isolation unit referral. the present analysis formalises the problem as a calibration-transfer evaluation and characterises the operating characteristics of a before-after-control-impact (BACAI) design that would estimate the effect of deploying such a model on referral timing. No observational data are used or claimed. Under a declared variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64, power was estimated across 2000 design cells defined by cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. The design requires 54 clusters to detect a 5 percentage point shift at 80% power. The analysis specifies the estimator, derives its variance, and reports the minimum detectable effect at 80% power for each cluster count. The contribution is a design-level account of what calibration transfer would have to change, and at what scale it could be detected, before an AI triage model can be said to alter isolation unit referral timing under variable case definitions.