Artificial Intelligence replication of District Hospital Antiretroviral Regimen Switching Under Unmeasured Viral Load Assay Turnaround Censoring: A Healthcare Survival-Robustness Evaluation

District hospitals in resource-limited settings switch patients between antiretroviral regimens on the basis of viral load results whose turnaround times are unmeasured and variable, so that the switching decision is made under a form of right-censoring that routine monitoring does not record. the present analysis treats that censoring as a structural feature of the switching problem rather than as a data defect, and asks how robust a regimen-switching rule can be when the delay between sample draw and result availability is itself unobserved. The work is a methodological and replication-based evaluation: a survival-analytic formulation of the switching decision is developed, an artificial-intelligence replication harness is specified, and the behaviour of the ordinary least squares estimator under a declared data-generating process is characterised as a calibration step for the harness. The evaluation drew 600 data sets of N = 500 from a process with intercept 12.0, exposure 0.42, group -3.1, severity 1.35 and error SD 4.2, using seed 160172507. Across the replications the intercept bias was 0.03959 with RMSE 0.86118, the exposure bias was -0.00077 with RMSE 0.01522, and nominal 95% coverage was 1.0 for both parameters. Turnaround censoring should be modelled explicitly as a latent hazard, that switching rules calibrated only on observed viral load will be systematically optimistic, and that replication-based robustness evaluation is a precondition for deploying such rules in district hospitals.