Scholarlink

<p>Patients with neuroendocrine tumors (NETs) often face substantial informational gaps and rely on the Internet for medical information despite variable quality and frequent misinformation. These challenges are particularly critical in complex care pathways such as radioligand therapy (RLT), where radiation safety constraints may increase anxiety and impact adherence. Large language models (LLMs) have emerged as promising tools for patient education, offering accessible and empathetic explanations tailored to patients’ level of understanding. However, their performance in specialized settings and across languages remains insufficiently evaluated. In a proof-of-concept study, five commercially available LLMs were evaluated by seven medical experts on 14 common patient questions about RLT for NETs in English and French, across three domains: accuracy, conciseness, and readability. LLMs demonstrated good performance, although significant differences were observed across models and languages. Performance was generally higher in English, highlighting cross-lingual disparities that may affect equitable access to reliable health information. While critical errors were uncommon, some responses illustrated how oversimplified information lacking clinical nuance may be misleading in institution-dependent contexts, particularly regarding hospitalization requirements. Readability analysis using the Flesch–Kincaid Grade Level indicated that responses often exceeded recommended levels for patient education. These findings suggest that LLMs could support patient education in specialized oncology settings but require careful integration into clinically supervised workflows, with potential advantages of domain-specific over general-purpose models. Improving readability, reducing cross-lingual disparities, and ensuring alignment with clinical practice will be essential to ensure safe, reliable, and equitable use of LLMs in patient education.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Large Language Models for Patient Education: Insights from Cross-Lingual Expert Evaluation in Radioligand Therapy for Neuroendocrine Tumors

  • 期刊论文

  • 发表日期：

    2026年4月28日

查看全文
  • Vincent Habouzit,
  • Alice Durand,
  • Laure Al Mansour,
  • Boumediene Lachachi,
  • Sarah Chaib,
  • Melissa Halilali,
  • Marc Janier,
  • Thomas Walter,
  • Thomas Grenier,
  • Anthime Flaus

摘要

Patients with neuroendocrine tumors (NETs) often face substantial informational gaps and rely on the Internet for medical information despite variable quality and frequent misinformation. These challenges are particularly critical in complex care pathways such as radioligand therapy (RLT), where radiation safety constraints may increase anxiety and impact adherence. Large language models (LLMs) have emerged as promising tools for patient education, offering accessible and empathetic explanations tailored to patients’ level of understanding. However, their performance in specialized settings and across languages remains insufficiently evaluated. In a proof-of-concept study, five commercially available LLMs were evaluated by seven medical experts on 14 common patient questions about RLT for NETs in English and French, across three domains: accuracy, conciseness, and readability. LLMs demonstrated good performance, although significant differences were observed across models and languages. Performance was generally higher in English, highlighting cross-lingual disparities that may affect equitable access to reliable health information. While critical errors were uncommon, some responses illustrated how oversimplified information lacking clinical nuance may be misleading in institution-dependent contexts, particularly regarding hospitalization requirements. Readability analysis using the Flesch–Kincaid Grade Level indicated that responses often exceeded recommended levels for patient education. These findings suggest that LLMs could support patient education in specialized oncology settings but require careful integration into clinically supervised workflows, with potential advantages of domain-specific over general-purpose models. Improving readability, reducing cross-lingual disparities, and ensuring alignment with clinical practice will be essential to ensure safe, reliable, and equitable use of LLMs in patient education.