Reading results
Orinyx splits each order into claims: a dose, a route and a frequency. Every claim gets one of four results. Each flagged or supported row cites the label section or monograph paragraph it was judged against.
| Result | What it means | What it does not mean |
|---|---|---|
| FLAGGED (with a severity) | The cited label text says otherwise. The severity comes from a curated rule for that kind of limit. | It does not say what the right order is. Orinyx flags; the clinician decides. |
| SUPPORTED | The claim agrees with the cited label text. | It is not a safety clearance, and not a judgment that the order suits a patient. It does not clear the daily total, the schedule or the patient context unless a row says so. |
| NOT VERIFIED (abstain) | Orinyx could not check this claim: no rule covers it, the order is ambiguous, or the evidence does not apply. The row says why. | It does not mean the claim is correct. An unchecked claim is unchecked. Treat it as needing review, never as passing. |
| UNREADABLE | The order could not be read as written, for example a drug name that is no drug, a unit that is no unit, or a body site given as a route. | Do not act on it. |
Severities are minor, moderate or major, from curated rules. There is no harm-weighted or critical tier.
UNREADABLE orders come first in the result, marked "do not act on it". They have no citation, because no label was consulted. Any suggestions shown are for display only, for the clinician to confirm.
NOT VERIFIED rows show no citation. This server prints - in the Citation column of every NOT VERIFIED row, even when the result names a specific label. The verdict is unchanged.
A run with no flags is not a clean bill of health. Every run says so: "This is not a safety clearance; claims not verified were not checked."
Check the citation. It shows which product a row was judged against. Check that the cited product matches the order's route, population and formulation. See Known limits.
PATIENT CONTEXT NOT EVALUATED
If the request mentions allergies or patient history, the result says PATIENT CONTEXT NOT EVALUATED. Order-text checks read the order only. Patient context and allergies are not evaluated.
Run ids
Each result ends with a Run id: line. The header shows a shortened id. get_verification, get_accuracy and export_verification need the full run id from the Run id: line.
Source: GitHub v0.1.2
Version: v0.1.2, updated 2026-11-03