Rural transfer distance absent from capacity models

1. System failure identification

   A rural patient is routed to a facility with an available bed two hours away, while a nearer facility with a longer queue would have delivered care sooner. Time to treatment rises while the routing decision records a success.

2. Data pipeline diagnostics

   Matching optimises on bed availability alone. Travel time, weather, and transport crew availability are not inputs, so the model minimises the wrong quantity. The omitted variables are precisely the ones that dominate in rural geography.

3. Predictive computational model

   Optimise expected time to definitive care: travel time plus projected wait on arrival. A nearer queue frequently wins on the combined figure. Both terms are estimable, one from routing data and one from queue state.

4. Workflow integration

   Present both components in the routing recommendation. Coordinators override on local knowledge the model lacks, and a composite score they cannot decompose will simply be distrusted. Transparency is what makes an override an informed decision rather than a rejection.

Back to the case study database