Guide-linked quiz · Medium Level
AI for Systematic Review Searching Quiz
Test AI systematic-review workflows: search reproducibility, active-learning screening, missed records, human review, extraction validation, and PRISMA
Question 1 of 8
What does an active-learning screening system typically do after a reviewer labels records?
Preview every question8 questions
Answers and explanations appear as you take the quiz.
- What does an active-learning screening system typically do after a reviewer labels records?
- Automatically declares all unshown records ineligible
- Uses those labels to prioritize the next records likely to be relevant
- Replaces the review protocol
- Generates the final evidence synthesis without source checking
- Why should teams check low-ranked records when using AI prioritization?
- Treat every low-ranked record as excluded once a set count is reached
- Stop when a number of consecutive irrelevant records is reached without checking recall
- Relevant studies may still be ranked low and could be missed by an early stop
- Assume the classifier’s confidence proves an article is ineligible
- Which automation detail does PRISMA 2020 ask reviewers to report when applicable?
- Only the model vendor’s name
- Only the final number of included papers
- The number of pages in each article
- How many reviewers screened records, whether they worked independently, and automation tools used
- A language model extracts an outcome value from a paper. What should the reviewer do?
- Verify the value against the source table or text and record its location
- Ask the same model to verify its own extracted value
- Compare only with an abstract-level summary, not the paper
- Infer the value from the article title and study population
- Why preserve exact database search strings and dates?
- So an AI model can change eligibility criteria
- So another team can reproduce or update the search
- To prove that all relevant studies were found
- To replace the protocol
- What does a stopping rule based on consecutive irrelevant records require?
- A guarantee that recall is 100 percent
- A replacement for full-text screening
- Assessing and reporting the risk that eligible records remain below the threshold
- No human validation
- How should AI-generated search synonyms be treated?
- As automatically valid for every database
- As a reason to skip controlled vocabulary
- As evidence that the search is comprehensive
- As candidates to test with domain expertise and document in the final strategy
- Who remains responsible for the review’s coverage and conclusions?
- The review authors
- The screening software vendor
- The database provider
- The model that ranks abstracts
Keep testing yourself
More quizzes picked for your level.