Bài kiểm tra có liên kết hướng dẫn · Trung bình Cấp độ

AI cho bài kiểm tra tìm kiếm đánh giá có hệ thống

Kiểm tra quy trình đánh giá hệ thống AI: khả năng tái tạo tìm kiếm, sàng lọc học tập tích cực, hồ sơ bị bỏ sót, đánh giá của con người, xác thực trích xuất và PRISMA

Đường dẫn hướng dẫn liên quanAI để tìm kiếm đánh giá có hệ thống
Câu hỏi 1 của 8

What does an active-learning screening system typically do after a reviewer labels records?

Xem trước mọi câu hỏi8 câu hỏi

Câu trả lời và giải thích xuất hiện khi bạn làm bài kiểm tra.

  1. 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
  2. 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
  3. 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
  4. 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
  5. 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
  6. 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
  7. 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
  8. 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