You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Improve mapping logic and robustness across CSV2FHIR maps.
While converting Observation/gram-stain from CSV to FHIR, it was observed that some component data was missed on mapping. In this instance, the bug was that the CodeableConcept is not created if coding is not present/populated in the CSV.
Expected behaviour is that CodeableConcept is created if any of the codings or text elements are populated.
A band-aid fix was made only in the CSV2Observation.map (see commit 7952765) to produce the expected output however in general, all the maps need review and improvement to mapping logic, consistency, and robustness.
The text was updated successfully, but these errors were encountered:
There is an issue with FHIR Mapping language and flat data structures.
FHIR Mapping language reusable component is group.
Group allows you to map one subtree element to another.
The group is invoked by providing two subtree source and destination.
This approach allows you to modularise mappings.
Here you can find an example that maps QuestionnaireResponse to Patient https://github.com/beda-software/FHIRPathMappingLanguage/blob/main/examples/repeatable/fhirmapping.map that uses this technique.
I tried to leverage it for the current project.
Unfortunately it will not work.
Since our source format is CSV, we don't have a subtrees, but it is a requirement for Group in FML to be invoked.
I am proposing to switch from CSV to QuestionnaireResponse format.
With QuestionnaireResponse data structure we can leverage groups and simplify mappings that enhance, readability, support on the long term and allow us avoid bugs like this one across all mappers. Since we can use group as a reusable element for many mappings.
Improve mapping logic and robustness across CSV2FHIR maps.
While converting Observation/gram-stain from CSV to FHIR, it was observed that some
component
data was missed on mapping. In this instance, the bug was that theCodeableConcept
is not created ifcoding
is not present/populated in the CSV.Expected behaviour is that
CodeableConcept
is created if any of thecodings
ortext
elements are populated.A band-aid fix was made only in the CSV2Observation.map (see commit 7952765) to produce the expected output however in general, all the maps need review and improvement to mapping logic, consistency, and robustness.
The text was updated successfully, but these errors were encountered: