Artificial Intelligence replication of District Hospital Malaria Rapid Diagnostic Test Batch Allocation Under Unmeasured Reader Optical Drift: A Healthcare Robustness-Constrained Evaluation

Rapid diagnostic tests for malaria in district hospitals are allocated in batches, yet the optical reader used to interpret test lines can drift in ways that are never measured. The framework couples a linear measurement-error model of reader drift to a constrained optimisation of batch assignment across patient groups, and it derives finite-sample bounds on the bias and mean squared error of the exposure coefficient that governs the drift-allocation interaction. A replication study of 600 data sets with 500 observations each was conducted under a declared data-generating process with intercept 12.0, exposure 0.42, group -3.1, severity 1.35 and error standard deviation 4.2. The ordinary least squares estimator was characterised across replications: estimated bias was 0.02242 for the intercept and -0.00065 for the exposure coefficient, root mean squared error was 0.85681 and 0.01531 respectively, and empirical coverage of the nominal 95% confidence intervals was 1.0 for both parameters. The framework proposes that batch allocation rules be screened against a drift-robustness criterion before deployment, and it identifies the conditions under which the exposure coefficient remains estimable despite unmeasured reader drift.