Systematic Review of Reporting guidelines for large language models used in healthcare research

Résumé fourni par la source

This systematic review aims to synthesize existing reporting guidelines for large language models (LLMs) in healthcare research and evaluate their adequacy in addressing gaps in transparency, reproducibility, and clinical applicability. A systematic search was conducted to identify relevant studies on reporting guidelines for LLMs used in healthcare research using the PubMed database. We included 18 studies focused on reporting guidelines for LLMs used in healthcare research. The studies primarily aimed to develop or evaluate reporting frameworks to improve transparency, reproducibility, and methodological rigor in LLM applications. Several studies focused on creating structured reporting checklists for LLM applications in healthcare. The Chatbot Assessment Reporting Tool (CHART) was developed across multiple studies. Similarly, TRIPOD-LLM extended the TRIPOD+AI framework with 19 main items and 50 subitems, emphasizing modular reporting for diverse LLM tasks. Ultimately, while existing reporting guidelines represent an important advancement toward standardizing LLM research, their long-term impact will rely on broad adoption and iterative refinement to meet the evolving challenges of artificial intelligence.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.