Artificial Intelligence replication of District Hospital Radiotherapy Linear Accelerator Maintenance Scheduling Under Unmeasured Dosimetry Phantom Calibration Drift: A Healthcare Fault-Tree Reliability Evaluation

The problem is that calibration drift accumulates silently between phantom audits, so maintenance decisions are made on observed fault logs that systematically under-represent the true hazard state. A fault-tree model is specified in which top-event probability depends on component failure rates, maintenance interval, and a latent drift process; the drift process is treated as an unmeasured confounder that biases naive reliability estimates. The framework is evaluated through a replication-based replication: 600 data sets of N = 500 were drawn from a declared data-generating process (intercept 12.0, exposure 0.42, group -3.1, severity 1.35, error SD 4.2), and the behaviour of the ordinary least squares estimator was characterised under the stated process. Across the 600 replications, estimated bias was 0.00818 for the intercept and -0.00026 for exposure; root mean squared error was 0.86646 and 0.01499 respectively; empirical coverage of the nominal 95% confidence intervals was 1.0 for both parameters. The framework proposes that maintenance scheduling should be indexed to drift-adjusted hazard rather than to observed fault counts, and identifies the conditions under which the unmeasured drift term renders observed-count scheduling non-conservative.