Artificial Intelligence replication of Community Clinic Insulin Cold-Chain Routing Under Unmeasured Solar Refrigerator Duty-Cycle Gaps: A Healthcare Robustness Evaluation

Solar direct-drive refrigerators are widely deployed for insulin storage in off-grid community clinics, yet their duty-cycle behaviour is rarely logged, so routing models that assume continuous temperature control may be systematically optimistic. The evaluation is replication-based: 600 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, and the behaviour of the estimator was characterised across replications. The estimated bias was -0.02587 for the intercept and 0.00047 for the exposure coefficient, with root mean squared errors of 0.89102 and 0.01592 respectively, and empirical coverage of the nominal 95% confidence intervals was 1.0 for both parameters. The results indicate that under a correctly specified linear surrogate the estimator is close to unbiased and that interval coverage is conservative in this design, which bounds what a routing model can be expected to recover when duty-cycle gaps are absent from the covariate set. The contribution is a formal statement of the conditions under which routing surrogates remain trustworthy, and a specification of the additional data a clinic network would need before routing decisions could be defended on thermal grounds.