August 25, 2025Open Access

Large Language Models for Cancer Communication: Evaluating Linguistic Quality, Safety, and Accessibility in Generative AI (Preprint)

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Mixed-methods evaluation reveals trade-offs in linguistic quality and safety of LLMs in cancer communication.

General-purpose large language models improved linguistic quality but were less accessible due to complexity,
Medical large language models generated simpler, clearer content, yet scored lower in safety and empathy,
Mixed-methods evaluation utilized quantitative and qualitative metrics to assess performance across models,
Future development of health-focused large language models must balance specialization and overall safety to improve trust and clarity.

Saha et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark