Methodological Adherence, Reporting Completeness, and Risk of Bias in AI-Assisted Systematic Reviews: A Matched Meta-Research Study

AI-assisted health systematic reviews published 2023–2025 scored 1.64 points lower on a modified 13-item AMSTAR-2 adherence scale than matched reviews with no reported AI use.

• Why it matters: Evidence is limited on whether completed AI-assisted systematic reviews maintain methodological rigor and transparent reporting compared with matched non-AI reviews.
• What they did: In a matched-cohort meta-research study of 231 English-language, peer-reviewed health systematic reviews, AI-assisted reviews were 1:2 matched to non-AI reviews and assessed using modified AMSTAR-2, PRISMA 2020, and ROBIS.
• The result: AI-assisted reviews had less complete PRISMA reporting and, among assessable ROBIS ratings, were more often low risk (75.0% vs 65.6%), while AI-use transparency was incomplete.

AI summary of the abstract · may contain errors · not medical advice · read the paper.