Artificial Intelligence replication of District Hospital Ventilator Allocation Under Unmeasured Oxygen Cylinder Manifold Pressure Decay: A Healthcare Safety-Constrained Reinforcement Learning Evaluation

Ventilator allocation in district hospitals is complicated by an unmeasured hazard: oxygen cylinder manifold pressure decays unpredictably, so a ventilator that appears available may deliver a hypoxic mixture within minutes. Because no field data on manifold decay are available, the framework is evaluated through a replication-based replication study of the underlying regression estimator that would calibrate the safety constraint. Six hundred data sets of N = 500 were drawn from a declared data-generating process with intercept 12.0, exposure 0.42, group -3.1, severity 1.35 and error SD 4.2, using seed 2055512084. The ordinary least squares estimator was characterised across replications. Estimated bias was 0.02661 for the intercept and -0.00031 for exposure; root mean squared error was 0.85648 and 0.01524 respectively; empirical coverage of the nominal 95% confidence intervals was 1.0 for both parameters. The results indicate that the estimator recovers the exposure coefficient with negligible bias and near-nominal coverage, supporting its use as the calibration layer of the safety constraint.