How do GPs Want Large Language Models to be Applied in Primary Care, and What Are Their Concerns? A Cross-Sectional...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Primary Care Technology

Background:

• Large language models (LLMs) offer significant potential in medicine, yet their specific application and associated concerns within primary care remain unexplored.
• General Practitioners' (GPs) perspectives are crucial for the successful integration of LLMs into clinical practice.

Purpose of the Study:

• To assess how GPs envision LLMs being utilized in primary care settings.
• To identify GPs' primary concerns regarding the implementation of LLMs in their clinical workflows.

Main Methods:

• An online survey was distributed to UK-based GPs via two specialized Facebook groups.
• The survey collected data from 107 practicing GPs between November 6-13, 2024, using non-probability sampling.

Main Results:

• GPs expressed strong support for LLMs performing both non-clinical and clinical tasks, assuming guaranteed accuracy and safety.
• The most favored non-clinical task was real-time consultation note generation (44.0%), while the top clinical task was automated identification and actioning of outstanding clinical tasks (51.0%).
• Patient safety emerged as the most significant concern (36.2%) among GPs regarding the integration of LLMs into clinical systems.

Conclusions:

• Preliminary evidence suggests significant potential for LLM utility in primary care, aligning with GP preferences for specific applications.
• Further research is essential to refine LLM development for primary care, ensuring technological advancement and user acceptance among GPs.