Artificial Intelligence Constraint-Aware Deep Reinforcement Learning for Emergency Blood Inventory Rotation Under Variable Cold-Chain Power Outage Telemetry: A computed Evaluation of Healthcare Wastage Equity

Emergency blood inventory rotation under intermittent cold-chain power telemetry is a sequential decision problem in which the cost of a wrong rotation is borne unequally across facilities. The agent is trained with a Lagrangian constraint on wastage disparity across facility tiers, so that the policy maximises expected usable-unit-days subject to an explicit equity budget. The evaluation characterises the operating characteristics of the design across cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points, with 2000 replications per cell. The design requires 54 clusters to detect a 5 percentage point effect at 80% power. Equity-constrained rotation policies are identifiable in principle but that the telemetry-missingness regime, not the learning algorithm, is the binding constraint on what a field evaluation can detect.