-
Notifications
You must be signed in to change notification settings - Fork 0
LLM
Sam Hume edited this page Aug 14, 2025
·
11 revisions
- Select variables needed for lab abnormality analysis
- Compare variable definitions against study XXX (multiple define.xmls)
- Check descriptions for grammar and logical issues
- Check compliance with CDISC TAUG
- Check that my Define-XML conforms ADaMIG X.X
- Check for circular dependencies in variable derivations
- Check compliance with FDA SDTCG (PMDA/EMA/NMPA/FDA BIMO guide)
- Check that all Define-XML business rules are satisfied
- Build a dependency tree of the datasets in define.xml
- Are there any variables missing, which are required by the standard (probably a CORE check and should be excluded)?
- Which dataset can support primary efficacy analysis? (SAP required as context)
- Which study-specific validation checks should I add? (Can be dataset/variable level question)
- Are there any study-specific conventions?
- Identify all datasets and variables which need to be updated to implement change XXX.
- Suggest improvements to the define.xml metadata definitions
- List all SDTM datasets used/not used in my ADaM derivations
- Check dataset is compliant with a standard/TAUG.
- Check all CRF fields from XX page/YY section are mapped in this dataset
- Explain the purpose of this dataset to me
- Identify non-standard things in this dataset (e.g. we derive additional age variable)
- Explain purposes for each of the ANLxxFL flags
- Check if variable names follow CDISC conventions
- Check origins
- Check if there are variables which should have codelists associated with them
- Check if VLM should be added for any variables
- Check variable names are following a consistent naming conventions
- Check all OIDs to make sure all references are present and no unused variables
- Which variables represent baseline characteristics?
- Which datasets contains efficacy data?
- Which variables depend on variable DS.VAR?
- Find variables with incomplete or vague descriptions
- Check if there are codelists which should be CDISC/NCI codelists
- Check if there are codes which should be from CDISC/NCI codelists instead of extended values (i.e. code values are synonyms and need to be changed)
- Check for inconsistent terminology usage across similar variables
- Identify complex derivation algorithms that need validation
- Check all analysis variables are listed
- What are SAP parts supporting this ARM output?
- Write a program in XXX language supporting this ARM output.
- Generate text for a DRG based on findings in the Define-XML
- Generate basic statistics describing how much of the Define-XML is based on standards
- Counts, such as how many datasets are in the Define-XML or how many variables have associated value lists
- Listing of define elements, such as datasets, variables, or value-level metadata