cdisc-org / cdisc-rules-engine

Open source offering of the cdisc rules engine
MIT License
49 stars 13 forks source link

As RULE ENGINE, I need to access a list of FDA-desired trial summary parameters (TSPARM) specified in their Technical Conformance Guide (TCG) so that I can perform this kind of FDA recommended data quality check. #759

Open chowsanthony opened 4 months ago

chowsanthony commented 4 months ago

Background:

  1. FDA owns and publishes the TCG.
  2. A new version of TCG automatically overrides its predecessor.
  3. As of v5.7 published March 2024, this information of desired TSPARM can be found in Appendices B and C. Two lists are available, separated by clinical and non-clinical submission use cases. This information is unstructured text in a PDF file format document.
  4. There are no known equivalent recommendations from PMDA.

Perceived use:

  1. CDISC will update and maintain the information followed each TCG revision.
  2. USER may amend the information to meet the uniqueness in each of their submissions. Ref: "Sponsors may use additional TS Parameters not listed in the table below if they are needed to describe the trial." TCG v5.7.
  3. CDISC will write one CORE rule, e.g., The TS dataset does not meet all the FDA desired parameters, as specified in, e.g., my_fdatcg_tsparm_spec.yaml.

Outstanding questions:

  1. Do we need to evaluate TSGRPID's role for each parameter listed in the Appendices?
  2. How do we manage the release of this resource in the future? Note that there is no set publication cadence for TCG revisions.

Draft acceptance criteria:

  1. A structured, machine-readable representation of information in said TCG appendices, especially conditions in the FDA Notes column.
  2. This information is a separate resource file, in YAML format.
  3. This resource file should contain an attribute that describes the TCG version.
  4. The engine will output error messages contextualized to each TSPARM, not a generic message.
  5. The engine will need to know which submission use case (e.g., clinical, non-clinical, tobacco) so that the correct set of parameter specifications is used.
mhungria commented 4 months ago

Perceived Use #3: I think it would be "CDISC will write a set of CORE rules, ...".

mhungria commented 4 months ago

Draft acceptance criteria #5: I think we need to add "Assumption: a Define-XML v2.1 document is expected for appropriate implementation of the TCG related rules.".

mhungria commented 4 months ago

Draft acceptance criteria #2: "This information is a separate resource file" could be stated in a more generic way "This information is a separate set of resource file(s)".

mhungria commented 4 months ago

I would add: Draft acceptance criteria #2.1: The engine will include functionality to handle this type of information resource as a reference resource where all its keys/value pairs could be queried by the user.

mhungria commented 4 months ago

Draft acceptance criteria #4: I would say that this criteria is not for the engine, instead it corresponds to how each rule is designed.

chowsanthony commented 3 months ago

We probably should and could apply this to Trial Set parameters (TXPARM). Excerpt from TCG:

The Trial Sets domain (TX) should be submitted for each study. Every set in the TX domain should have only one record with each of the following TXPARMCD values: SPGRPCD (sponsor group code associated with the set), GRPLBL (sponsor group label associated with the set), PLANMSUB (planned number of males in set), and PLANFSUB (planned number of females in set). There should be a one-to-one correspondence between GRPLBL and SPGRPCD entries in the TX domain.