Biostatisticians Meet AI : Navigating Shifts While Preserving Principles

作者：Bin Zhu · 发表于：Statistics in Medicine · 年份：2025 · DOI：10.1002/sim.70271 · 被引用次数：1 · 研究领域：Statistics Education and Methodologies

The manuscript ChatGPT as a Tool for Biostatisticians: A Tutorial on Applications, Opportunities, and Limitations by Dobler et al. provides a timely and insightful assessment of how large language models (LLMs) can assist in biostatistical work. The authors showcase several real-world use cases to illustrate ChatGPT's capabilities and pitfalls in our domain. This breadth of examples, each repeated in multiple ChatGPT sessions to assess consistency, offers a nuanced picture of what current LLMs can and cannot do. The effort is commendable for moving beyond abstract discussions and demonstrating concrete applications of ChatGPT in biostatistical workflows. Notably, the tutorial highlights instances of impressive performance (e.g., successful latent class analysis) alongside clear failures (e.g., incorrect meta-analyses), thereby educating biostatisticians on both the opportunities and limitations. The authors' ultimate message is one of cautious optimism: LLMs can streamline routine tasks for biostatisticians, but their outputs must be verified by a knowledgeable user. By openly sharing both positive results and errors, Dobler et al. set a pragmatic tone and reinforce that LLMs, while transformative, are no substitute for sound statistical judgment. We applaud the authors for this balanced, real-world tutorial, which serves as a starting point for biostatisticians interested in harnessing artificial intelligence (AI) in practice. The way we conduct statistical analysis has cont...