Large Language Models in Medical Diagnostics: Scoping Review With Bibliometric Analysis

Area of Science:

• Medical Diagnostics
• Artificial Intelligence
• Health Informatics

Background:

• Large language models (LLMs) offer potential to improve medical diagnostics and address healthcare disparities.
• Rapid LLM advancements necessitate a synthesis of their applications, challenges, and future directions in medicine.

Purpose of the Study:

• To provide a comprehensive overview of current research on LLMs in medical diagnostics.
• To identify commonly used LLMs, assessment methods, performance, and medical domains involved.

Main Methods:

• Scoping review following Joanna Briggs Institute Manual and PRISMA-ScR guidelines.
• Literature search across major databases (2022-2025) with bibliometric analysis using VOSviewer.
• Data extraction on LLM types, applications, and performance metrics.

Main Results:

• Rapid growth in LLM research, dominated by GPT-4 and GPT-3.5.
• Key applications include disease classification, medical question answering, and diagnostic content generation.
• High accuracy in radiology, psychiatry, neurology, but noted biases and ethical concerns (privacy, hallucination) hinder adoption.

Conclusions:

• LLMs have transformative potential in diagnostics, requiring validation, bias mitigation, and multimodal integration.
• Future research should focus on explainable AI, specialty optimization, and regulatory harmonization for safe deployment.