Kit 7 · MedCheck · deliverable · designed and built by I Wayan Usada · fictional case, not clinical guidance
In this fictional brief: fewer routine-looking dispensing errors reach the ward, without a rise in unnecessary prescriber queries. Proposed measures are on the case page; nothing has been measured.
First-year technicians check the drugs they know are dangerous and skim the routine ones. Some compensate by querying everything, which slows the dispensary and teaches prescribers to ignore queries.
| # | Behaviour | What it looks like | Where the build practises it |
|---|---|---|---|
| 1 | Turn each directive into a week | count tablets per day and per slot before signing | 3D "Fill the week" (required); model annotations in 2D |
| 2 | Read the chart line, not only the prescription | allergies and diagnosis checked against every line | Bakker: the allergy is only on the chart |
| 3 | Name the line and the kind of problem | a query that says what and why | line + category scored separately |
| 4 | Clear clean prescriptions with confidence | no defensive queries | Visser: any flag scores 0 and costs audit minutes |
| 5 | Write a usable annotation | drug, reason, next step | free-text annotation compared with a model answer |
It will not teach pharmacology, and it cannot fix workload or interruptions at the dispensing bench. Those are named on the case page as system factors.
Up to 3 per prescription, 15 in total; pass = 12 or more with no critical error released. Annotations are self-compared, not auto-scored.
Prefer the source? Download the raw markdown.