Review: adam-adqs-missed-visit-locf

[wangben718]

Review of the adam-adqs-missed-visit-locf benchmark.

Question	Answer
Task: Would you agree that the benchmark use case represents real data derivation in clinical trials?	Disagree
Inputs: Would you agree that the inputs and outputs represent real data structures collected in clinical trials?	Disagree
AI-assisted review: Would you agree with how this benchmark handles each potential concern below?	Disagree
Overall	Not approve

Comments

I believe that this need another look. 1. Baseline value is carried forward in the example, e.g, for subject 103. Is this a common practice or only for Alzheimer. 2. The description of this example sounds like a Alzheimer-specific case. 3.input qs have several ADaM variables. 4. The collected Week 8 record for subject 103 with an empty score is dropped. In practice that record usually stays (AVAL missing, traceable to its QSSEQ), and the LOCF value goes on a separate derived record.
5. The added Week 8 record for subject 104 has DTYPE = LOCF and no value. A derived record with nothing carried is questionable.

[pzhang724]

@wangben718 Thank you for this review. We changed adam-adqs-missed-visit-locf based on it (#4367, merged in #4399). The website shows the new version now.

Please review it again:

1. Open the benchmark page.
2. Answer the questions and select Submit and next.
3. On GitHub, select Start discussion.

This makes a new review. Your new review replaces your old review. Please do not edit your old review, and do not reply to it with new answers. A reply does not count as a review.

Thank you!