用生成式AI提速文献综述，同时保证可复现性与透明度。
L-PRISMA: An Extension of PRISMA in the Era of Generative Artificial Intelligence (GenAI)
- 人类主导+统计预筛选，结合AI提升效率
- 保持可复现性，避免大模型幻觉与偏差
- 适合需要高效且严谨的综述研究者
系统评价与元分析首选报告项目（PRISMA）框架为证据综合提供了坚实基础，但数据提取和文献筛选仍依赖人工，耗时且受限。近年来生成式人工智能（GenAI），特别是大语言模型（LLMs），为自动化与规模化这些任务带来机遇，从而提升效率。然而，LLMs固有的非确定性及幻觉、偏见放大的风险，正威胁着PRISMA的核心原则——可复现性、透明性与可审计性。为此，本研究将人类主导的综合与生成式AI辅助的统计预筛选相结合：人类监督确保科学有效性与透明度，而确定性的统计层则增强可复现性。所提方法系统性地扩展了PRISMA指南，为在系统评价流程中负责任地引入GenAI提供了可行路径。
原文摘要 · Abstract (English)
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework provides a rigorous foundation for evidence synthesis, yet the manual processes of data extraction and literature screening remain time-consuming and restrictive. Recent advances in Generative Artificial Intelligence (GenAI), particularly large language models (LLMs), offer opportunities to automate and scale these tasks, thereby improving time and efficiency. However, reproducibility, transparency, and auditability, the core PRISMA principles, are being challenged by the inherent non-determinism of LLMs and the risks of hallucination and bias amplification. To address these limitations, this study integrates human-led synthesis with a GenAI-assisted statistical pre-screening step. Human oversight ensures scientific validity and transparency, while the deterministic nature of the statistical layer enhances reproducibility. The proposed approach systematically enhances PRISMA guidelines, providing a responsible pathway for incorporating GenAI into systematic review workflows.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。