Large Language Models in Lung Cancer: Systematic Review

Area of Science:

• Medical Informatics
• Artificial Intelligence in Oncology

Background:

• Artificial intelligence (AI) and large language models (LLMs) are increasingly utilized in healthcare.
• LLMs offer advanced capabilities for complex tasks and interactive features in medical applications.

Purpose of the Study:

• To systematically review current applications of LLMs in lung cancer (LC) care.
• To evaluate the potential of LLMs across the full spectrum of LC management.

Main Methods:

• A comprehensive literature search was conducted across 6 databases up to January 1, 2025, adhering to PRISMA guidelines.
• Included studies were journal articles, conference papers, and preprints reporting LLM content in LC with original, separately presented LC data.
• Quality assessment used established tools (e.g., QUADAS-2, PROBAST, RoBINS-I), with data extraction focusing on model type, application, prompts, I/O, outcomes, and safety.

Main Results:

• Twenty-eight studies published between 2023-2024 were included, demonstrating LLM capabilities in medical record extraction, LC knowledge dissemination, and clinical decision support.
• Emerging visual and multimodal functionalities of LLMs were noted.
• Prompt engineering varied from zero-shot to fine-tuned methods; quality assessment indicated acceptable rigor but highlighted areas for improvement in bias control and data security.

Conclusions:

• LLMs present significant potential to enhance lung cancer diagnosis, patient communication, and clinical decision-making.
• Responsible integration necessitates careful consideration of data privacy, model interpretability, and the importance of human oversight.