Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Leveraging Large Language Models for Systematic Reviews - a Replication Study and Frame-work Proposal
ABSTRACT
Background:
Systematic reviews are resource-intensive, and Large Language Models (LLMs) have emerged as promising tools for systematic reviews; however, their ability to replicate complete systematic review processes has been insufficiently evaluated.
Objective:
To determine whether LLMs can accurately replicate key stages of a published systematic review, we assessed performance across the review process and developed a structured framework for transparent and reproducible LLM-assisted systematic reviews.
Methods:
By leveraging Application Programming Interfaces (API), two LLMs (Gemini-2.5-Pro and OpenAI o3) classified titles/abstracts and full texts, and subsequently extracted data from eligible articles. To achieve optimal performance, we iteratively refined the prompt and evalu-ated the classification on inclusion/exclusion criteria level.
Results:
Combined, the two LLMs reached an NPV of 0.99 and FNR of 0.07 with Cohen’s κ of 0.94 (95%-CI=[0.91-0.96]) for title and abstract, and an NPV of 0.95 and FNR of 0.04 with κ of 0.94 (95%-CI=[0.89-0.99]) for full text classification. LLM-based data extraction revealed an overall exact match score of 86.75% after semantic discrepancy resolution.
Conclusions:
Considering the successful replication and the strong agreement after ground-truth reconciliation, we propose the AIMS framework and the llmreviewR software as accessible co-screening tools for applying LLMs in systematic reviews.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.