Map Target Provenance Extension¶
| Canonical | https://fhir.apps.health/StructureDefinition/nexus-map-target-provenance |
|---|---|
| Status | draft · 1.33.0 |
| Base | Extension (constraint) |
| Context | ConceptMap.group.element.target (element) |
| Source | FSH · JSON |
Where a ConceptMap target came from and how confident its source is in it: provenance names the
source, and confidence is that source's confidence, a decimal from 0 to 1.
Overview
Confidence is comparable across sources, by construction. A machine classifier states its own
probability that the target belongs to the source code's related family. A rule-based source states
a fixed confidence no higher than the lowest a classifier target can carry, so sorting targets by
confidence never ranks a rule-based target above a classifier one. Confidence is stated to two
decimals and rounded down, never up.
It is not the relationship. target.equivalence still says how the target relates to the source
code; this says who proposed it and how sure they are. A confident target can still be relatedto.
The same facts may also lead target.comment, as [<provenance> p=<confidence>] (for example
[jev p=0.93]). Where both are present they agree; this extension is the structured form, and a
consumer should read it in preference to parsing the comment.
Terminology bindings
| Path | Strength | Value set |
|---|---|---|
Extension.extension.value[x] |
extensible | Map Target Provenance Kind |