Artificial Intelligence Triage for Snakebite Envenoming Under Variable Antivenom Stock-Out Conditions: A Constraint-Aware Evaluation of Referral Timing Errors Across Rural Health Centres

Antivenom stock-outs are a structural feature of rural snakebite care, yet triage algorithms are typically evaluated as if antivenom were always available. The framework formalises triage as a constrained sequential decision problem in which the triage agent must choose between treating locally, referring immediately, or observing, subject to a stochastic antivenom supply process and a patient-condition process that evolves between decision epochs. Two error types are defined: false-localisation errors, in which a patient who requires antivenom is retained at a facility that cannot supply it, and false-referral errors, in which a patient who could have been managed locally is referred, consuming scarce transport and receiving-capacity resources. A replication study was conducted to characterise the operating characteristics of the framework 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 for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. The analysis indicates that 56 clusters are required to detect a 5 percentage point effect at 80 percent power. The framework's principal contribution is to make the stock-out constraint explicit in the triage objective, so that referral timing errors are evaluated against the supply conditions under which they occur rather than against an idealised antivenom-available baseline.