AI disclosure for systematic reviews

Copy-paste methods text for reporting AIPRA use—aligned with the 2025 Cochrane/Campbell/JBI/CEE position statement and RAISE expectations for transparency and human oversight.

Journal editors and methodologists increasingly require authors to name AI tools, describe which stages used assistance, cite validation evidence where available, and document human checking. Cochrane’s Screen4Me documentation models this with boilerplate authors can paste into manuscripts. The paragraph below follows that pattern for AIPRA.

Adapt the paragraph to your protocol: remove stages you did not use, add exact software version/date, and report quantitative agreement rates if you measured them.

Methods paragraph

AIPRA (Artificial Intelligence Powered Research Automation; https://aipra.co) was used to support selected stages of this systematic review, including research-question framing, search organization, title/abstract and/or full-text screening recommendations, structured data extraction support, evidence-synthesis planning, and/or manuscript drafting, as applicable to this project. Outputs were generated with human oversight: reviewers verified AI suggestions against protocol-defined eligibility criteria and source records before accepting inclusion, exclusion, extraction, or manuscript text. Tool name, version/date of use, and the stages where AI assistance was applied are documented in our protocol and review records. Use of AIPRA for this workflow is supported by a published non-inferiority evaluation comparing an AIPRA-assisted pipeline with a human-led pipeline on the same review question (medRxiv preprint: https://www.medrxiv.org/content/10.1101/2025.10.27.25338904v1), which found overall quality meeting a pre-specified non-inferiority criterion while noting higher human scores on screening and extraction domains and emphasizing the need for human oversight. Reporting of AI use follows the expectations of the 2025 joint position statement on AI in evidence synthesis (Cochrane, Campbell Collaboration, JBI, and Collaboration for Environmental Evidence) and the RAISE recommendations for transparent tool selection and reporting.

Checklist before you submit

  • Tool name and URL (AIPRA / aipra.co)
  • Version or date range of use
  • Which workflow stages used AI assistance
  • Human oversight strategy (who verified what)
  • Citation to validation evidence — see research & validation
  • Any protocol deviations related to AI use

Primary governance sources: 2025 joint AI position statement (PMC); RAISE materials on OSF. Current PRISMA reporting standard for reviews is PRISMA 2020 (prisma-statement.org); PRISMA-AI remains under development on EQUATOR.