Applications of Large Language Models in Medical Research: From Systematic Reviews to Clinical Studies

Area of Science:

• Medical Research
• Artificial Intelligence
• Scientific Writing

Background:

• Large Language Models (LLMs) are increasingly integrated into medical research.
• Their impact spans systematic reviews, scientific writing, and clinical research workflows.

Purpose of the Study:

• To synthesize current evidence on the applications of LLMs in medical research.
• To evaluate the efficacy and limitations of LLMs in various research contexts.

Main Methods:

• A narrative review of literature published between 2023-2025.
• Searches conducted across major scientific databases (PubMed, Scopus, Web of Science, arXiv, medRxiv, Google Scholar).
• Inclusion of studies with empirical findings or methodological evaluations of LLM applications.

Main Results:

• LLMs demonstrate high accuracy (80-94%) in systematic review data extraction but show limited agreement (κ = 0.16-0.43) in risk-of-bias assessment.
• Scientific writing applications exhibit significant hallucination rates (47-55%) and demographic bias (>90%), requiring stringent verification.
• LLMs assist in clinical research tasks like coding and protocol development, but human validation is essential.

Conclusions:

• LLMs are powerful yet unstable tools in medical research, demanding continuous human verification.
• Maintaining a human-in-the-loop approach is crucial to balance AI efficiency with critical thinking and prevent cognitive offloading.