ESTRO 2025 - Abstract Book

S2186

Interdisciplinary – Education in radiation oncology

ESTRO 2025

Table 2: Characteristic and results of meta-analyses.

Conclusion: While PE is effective in enhancing pain knowledge and medication adherence, its impact on pain relief remains inconsistent, with evidence suggesting that intensity and duration of the educational intervention play a critical role. The review highlights the need for further research to identify cost-effective and optimal PE strategies, particularly in specific oncological contexts, to enhance patient outcomes and guide clinical practice.

Keywords: oncological patients, pain education

2869

Digital Poster Differences in Cancer Treatment Information by English/Spanish Bilingual Prompting of a Large Language Model Chatbot Javier Mora 1 , Shan Chen 2 , Raymond H Mak 2 , Danielle S Bitterman 2 1 Department of Radiation Oncology, Mass General Brigham, Boston, USA. 2 Department of Radiation Oncology, Brigham and Women’s Hospital/Dana-Farber Cancer Institute, Boston, USA Purpose/Objective: Large language model (LLM) chatbots can be a potential tool for patient health literacy and self-education, yet LLMs are predominantly trained on English content. It remains understudied whether chatbot responses vary depending on user language. We assessed LLM performance in providing cancer treatment information in English and Spanish. Material/Methods: A set of 104 questions about treatments for prostate, breast, and lung cancers of varied disease extent were written in English and translated by a bilingual oncologist into Spanish. GPT3.5-turbo-0613 was prompted. Quality concordance of chatbot responses were scored by a bilingual oncologist using U.S. National Comprehensive Cancer Network 2021 guidelines (based onGPT’s knowledge cut-off at that time) and pre-specified criteria (Table). Mann-