AI Systematic Review PRISMA-P Protocol Narrative: Drafting Eligibility and Search Summaries

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-research-AI-and-systematic-review-protocol-narrative-r8a3-creators

1. When using AI to draft a PRISMA-P protocol narrative, what is the primary value it provides?

   1. Restructuring raw notes into a coherent, decision-ready summary
   2. Determining which stakeholders should review the document
   3. Replacing the need for systematic review training
   4. Generating the final registered protocol without human input

2. A researcher asks AI to draft a PRISMA-P protocol narrative from their scattered notes. What should the AI output include to be useful for decision-making?

   1. A decision-ready summary with headline framing, substantive points with caveats, and explicit reviewer decisions
   2. A complete protocol ready for immediate registration
   3. A list of all studies found in preliminary searches
   4. A final risk-of-bias assessment for included studies

3. An AI generates a search strategy that only uses keyword searches without controlled vocabulary terms. Why is this problematic?

   1. Controlled vocabulary is only needed for qualitative reviews, not systematic reviews
   2. Keywords and controlled vocabulary are interchangeable in systematic review searches
   3. Keywords are more expensive to implement than controlled vocabulary
   4. Controlled vocabulary translation is required to capture all relevant studies; without it, about half of relevant studies are missed

4. Which of the following tasks is beyond AI's current capability when drafting PRISMA-P protocol narratives?

   1. Restructuring raw notes into coherent prose
   2. Summarizing PICO elements from the input notes
   3. Deciding which stakeholders need separate conversations before the document lands
   4. Identifying unresolved questions implied by the inputs

5. Before registering a PRISMA-P protocol that AI helped draft, what essential step must be completed?

   1. Submit directly to the registry since AI verified accuracy
   2. Delete all AI-generated content and rewrite manually
   3. Have a librarian review the search strategy
   4. Obtain approval from every stakeholder in the organization

6. An AI produces a protocol draft that omits any discussion of risk-of-bias assessment tools. What should the human reviewer do?

   1. Flag this as a gap—risk-of-bias tools must be specified in PRISMA-P protocols
   2. Approve it anyway since AI is generally accurate
   3. Submit to the registry and let reviewers catch the omission
   4. Request that AI add a brief mention without specifics

7. When AI flags unresolved questions in a draft protocol, what is the appropriate human response?

   1. Ignore them since AI drafting is usually accurate
   2. Accept the first resolution the AI proposes
   3. Delete the draft and start over without AI
   4. Address these questions explicitly before sign-off

8. A student uses AI to draft a PRISMA-P protocol and receives a polished narrative. The student wants to submit it for registration immediately. What is the main concern?

   1. PRISMA-P protocols cannot be registered online
   2. AI-generated content cannot be submitted to registries
   3. The protocol may lack explicit decisions the reviewer must resolve
   4. The student did not pay for the premium AI version

9. Why might AI struggle with 'reading the room' in the context of systematic review protocols?

   1. AI cannot access emotional tone in text
   2. AI always agrees with the first author
   3. AI is not programmed to recognize academic hierarchy
   4. AI cannot interpret political, ethical, or relational concerns that exist outside the analytical content

10. What distinguishes a high-quality AI-drafted protocol narrative from a poor one?

    1. Whether it includes explicit caveats and decision points for human resolution
    2. Whether it uses fancy vocabulary
    3. The length—longer drafts are always better
    4. Whether it was drafted on a weekday

11. A researcher realizes the AI-generated search strategy omits relevant database-specific subject headings. What is the most appropriate action?

    1. Remove the search strategy section entirely
    2. Submit as-is since the AI tried its best
    3. Use only Google Scholar going forward
    4. Consult a librarian to translate keywords to controlled vocabulary terms

12. When drafting eligibility criteria in a PRISMA-P protocol, what should the human ensure that AI has properly addressed?

    1. Clear PICO-aligned inclusion and exclusion criteria
    2. The exact number of studies that will be included
    3. The final list of included studies
    4. Statistical analysis plan for each study

13. An AI draft includes a perfectly written search strategy but fails to explain the rationale for database selection. What is the concern?

    1. AI is not allowed to mention databases
    2. Database rationale is optional in PRISMA-P
    3. Rationale must be determined by the funding agency only
    4. The draft lacks the substantive points with caveats required for decision-ready protocols

14. Why is stakeholder identification a task that must remain with humans, not AI?

    1. Determining who needs to review a document requires understanding organizational relationships, power dynamics, and communication needs that exist outside the analytical content
    2. Stakeholders must be listed alphabetically
    3. AI is not advanced enough to read names
    4. AI will always recommend too many reviewers

15. What is the relationship between AI-assisted drafting and librarian expertise in systematic review protocol development?

    1. AI can draft the protocol, but librarian review of search strategy is essential before registration
    2. AI should review librarian work for accuracy
    3. Librarians should not be involved until after registration
    4. Librarians are unnecessary when using AI