The TRIPOD-LLM reporting guideline for studies using large language models

TL;DR

TRIPOD-LLM is a checklist-based guideline to improve the reporting quality of LLM research in healthcare, emphasizing transparency and reproducibility.

Problem

LLMs are rapidly being adopted in healthcare, but the lack of standardized reporting guidelines undermines the reliability and reproducibility of study results.

Approach

Extending the existing TRIPOD+AI statement, we added LLM-specific items and developed 19 main items and 50 sub-items through a Delphi process and expert consensus.

Results & Contribution

We established a reporting standard for healthcare LLM studies, enhancing research quality and clinical applicability, and provide a website for guideline completion and PDF generation.