A Human-AI Collaborative Framework for Systematic Reviews

Li, Yingyin; Li, Jiao

Permalink

https://hdl.handle.net/2142/132982 Copy

Description

Title

A Human-AI Collaborative Framework for Systematic Reviews

Author(s)

• Li, Yingyin
• Li, Jiao

Issue Date

2026-03-12

Keyword(s)

• Systematic review
• Human–AI collaboration
• Large language models
• Evidence-based medicine

Abstract

Systematic reviews (SRs) are a cornerstone of evidence-based medicine (EBM), yet traditional workflows are increasingly unsustainable due to the exponential growth of medical literature and the labor-intensive nature of manual appraisal. While Large Language Models (LLMs) offer significant automation potential, their integration into clinical synthesis is limited by reasoning hallucinations and a lack of procedural transparency in complex critical judgment tasks. This paper proposes a holistic, five-stage human–AI collaborative framework designed to support the entire SR lifecycle. The core innovation is the formalization of a tri-modal interaction engine comprising Evidence-Grounded, Logic-Augmented, and Consensus-Mediated modes, which dynamically allocates cognitive tasks between human experts and AI. Validated through a large-scale risk-of-bias case study involving 2,100 RCT samples across seven bias domains, the framework demonstrates that structured human-AI synergy can stabilize judgment consistency, mitigate logical errors, and balance resource allocation with scientific rigor.. This work provides a strategic blueprint for the transition of information professionals into designers and managers of intelligent research systems.

Publisher

iSchools

Series/Report Name or Number

iConference 2026 Proceedings

Type of Resource

Other

Genre of Resource

Conference Poster

Language

eng

Permalink

https://hdl.handle.net/2142/132982

Copyright and License Information

Copyright 2026 is held by Yingyin Li and Jiao Li. Copyright permissions, when appropriate, must be obtained directly from the authors.

Owning Collections