Closed BSnoeijerCD closed 5 months ago
ElementId not added to test data Excel for studyCell while it is in the model. This involves cross class checking (since the elements are nested in studyDesign and only cross-referenced in studyCell). It is not possible to create this rule based on the current capabilities.
An elementId
column is not included in the StudyCell
sheet in the test data template because there can be more than one elementId
referenced from a StudyCell
. According to the current design of the test data template, the elementId
values would be listed in the Multiple_Values
sheet with parent_entity = "StudyCell"
and parent_rel = "elementId"
. However, this will not be the same as the converted USDM data, where the elementId
values will be listed in the string
dataset (with parent_entity
and parent_rel
populated as above), and there will also be a record in the StudyElement
table (with rel_type = "reference"
and with parent_entity
and parent_rel
populated as above) for each referenced elementId
.
This rule may not be possible even with the new cross-class joining because it might also need new "precondition" engine functionality - either for Match Datasets
or for the distinct
operation.
Given that, according to the model, the only way a StudyElement
can be referenced is from a StudyCell
, I think this check could be implemented by getting a distinct
list of rel_type
within id
from StudyElement
and then reporting where this list does not contain "reference" (i.e., a StudyElement
is defined but not referenced).
This only checks that each StudyElement
is referenced. There should also be a check to make sure that a StudyCell
should only reference StudyElement
s from the same StudyDesign
.
This is ready for review. The negative and positive test data is the same, except that the negative test data includes a definition of a "High - Start" element that is not referenced. Only one negative is expected.
@BSnoeijerCD @DianeWold Should there be an additional check/rule to ensure that each StudyCell
must only reference StudyElement
s from the same StudyDesign
(and similar checks for referenced StudyEpoch
s and StudyArm
s)?
Checks that a StudyCell only references StudyElements, StudyEpochs, and StudyArms from the same StudyDesign would make sense. I can define a rule for this; it could be split up if necessary.
Negative test data included StudyElement_4, which was not in any StudyCell, resulting in one negative result. Positive test data included five StudyElements, each of which was included in at least one StudyCell, resulting in one positive result.
Checks that a StudyCell only references StudyElements, StudyEpochs, and StudyArms from the same StudyDesign would make sense. I can define a rule for this; it could be split up if necessary.
I think it is good, based on this logic to split it up indeed.
Saving POC definition before overwriting for DDF4:
Authorities:
- Organization: 'CDISC'
Standards:
- Name: USDM
References:
- Citations:
- Cited Guidance: 'USDM'
Document: 'USDM v2.6'
- Cited Guidance: 'SDTMIG'
Document: 'SDTMIG v3.4'
Section: '7.2'
Origin: USDM Conformance Rules
Release Notes: ''
Rule Identifier:
Id: 'DDF00040'
Version: '1'
Validator Rule Message: ''
Version: '1.0'
Version: '3.0'
Check:
all:
- name: $element_rel_types
operator: does_not_contain
value: "reference"
Core:
Id: "CORE-000449"
Status: Draft
Version: '1'
Description: 'Each StudyElement must be linked to at least one StudyCell'
Executability: Fully Executable
Operations:
- group:
- id
id: $element_rel_types
name: rel_type
operator: distinct
Outcome:
Message: 'A StudyElement is not related to any StudyCell'
Output Variables:
- parent_entity
- parent_id
- id
- name
- label
- description
Rule Type: Record Data
Scope:
Entities:
Include:
- StudyElement
Sensitivity: Record
Create rule defined for #144