European Society of Clinical Microbiology and Infectious Diseases Study Group for Artificial Intelligence and Digitalisation consensus statements on key aspects in the interaction of clinical microbiologists and infectious disease specialists with general-purpose large language model-based systems: project protocol.

Abstract

General-purpose large language model (LLM)-based systems are increasingly accessible to clinicians and are being explored for applications in clinical microbiology and infectious diseases (ID). However, rapid adoption has outpaced the development of specialty-specific guidance, raising concerns related to safety, reliability, accountability, and antimicrobial stewardship. The project aims to develop consensus-based statements, endorsed by Study Group for Artificial Intelligence and Digitalisation of the European Society of Clinical Microbiology and Infectious Diseases (ESGAID), that describe principles, opportunities, and limitations in the interactions of clinical microbiologists and ID specialists with general-purpose LLM-based systems. Secondary objectives are to quantify expert agreement and identify areas of uncertainty and disagreement. The project follows a structured expert consensus design using the RAND/UCLA Appropriateness Method. A multidisciplinary panel of 15 experts will be selected through ESGAID using predefined criteria to ensure balanced expertise across clinical microbiology, infectious diseases, ethics, legal aspects, and patient safety. Ten draft statements, each supported by a structured literature review, will be developed by project coordinators. Statements will be evaluated through iterative rounds of anonymous rating on a 1-9 scale, combined with moderated remote discussions. Median scores will classify statements as supported, uncertain, or unsupported. Consensus will be defined as ⩾70% agreement with <15% disagreement during final anonymous voting. This protocol provides a transparent and reproducible framework to generate interim, specialty-specific statements on principles, opportunities, and limitations in the interactions of clinical microbiologists and ID specialists with general-purpose LLM-based systems. By combining structured evidence review with expert judgment, the resulting statements aim to delineate guidance on principles for interacting with these systems, highlight the nature of both existing risks and excessive skepticism, and identify research priorities in a rapidly evolving technological and regulatory landscape. Interactions with AI systems in microbiology and infectious diseases: an expert consensus project. Artificial intelligence systems such as ChatGPT, Gemini, or Claude are increasingly accessed by clinical microbiologists and infectious diseases specialists to explore medical questions or summarize information. However, clear understanding and deep knowledge about key principles, opportunities, limitations, and safeguards in the interaction with these tools may still be uncommon. For example, these systems can sometimes produce convincing but incorrect answers or lead users to rely too heavily on automated outputs. This project brings together experts to develop agreed statements on the principles guiding how clinicians can interact with these systems responsibly.