ESTRO 2025 - Abstract Book

S2200

Interdisciplinary – Education in radiation oncology

ESTRO 2025

4137

Digital Poster ChatGPT's Impact on Pre-therapy Patient Insight: Evaluating Experiences after Meningioma Radiation Treatment Diana-Coralia Dehelean 1 , Sebastian H. Maier 1,2 , Alev Altay-Langguth 1 , Alexander Nitschmann 1 , Michael Schmeling 1 , Daniel F. Fleischmann 1,3,4 , Paul Rogowski 1 , Christian Trapp 1 , Stefanie Corradini 1 , Claus Belka 1,2 , Stefan Schönecker 1 , Sebastian N. Marschner 1 1 Department of Radiation Oncology, University Hospital LMU, Munich, Germany. 2 Bavarian, Bavarian Research Center (BZKF), Munich, Germany. 3 German, German Cancer Consortium (DKTK), Munich, Germany. 4 German Cancer Research Center (DKFZ), German Cancer Research Center (DKFZ), Heidelberg, Germany Purpose/Objective: The study explores the potential of ChatGPT-4, an advanced large language model (LLM) by OpenAI, in educating patients about meningioma, a common type of brain tumor. Focusing on patients who have undergone radiation therapy, our analysis evaluates both assessments by clinicians and patient feedback to understand how the information generated by ChatGPT-4 is perceived. Material/Methods: A questionnaire was created based on responses generated by ChatGPT-4 to the following questions: 1. What are meningiomas? 2. What are treatment options for meningiomas? 3. What are the benefits of radiotherapy for meningiomas? 4. How is radiotherapy administered to the brain area? 5. What is stereotactic radiotherapy? 6. What are typical side effects of meningioma radiation? 7. What should you pay attention to during radiation therapy to the head? 8. What should be considered after radiation treatment? The final questionnaire was reviewed by nine experienced clinicians who evaluated the accuracy, relevance and completeness of the responses generated by ChatGPT-4. It was then administered to 26 patients during their follow up visit after radiation therapy. Items were rated on a 5-point Likert scale. Results: The study included 26 participants, aged 40 to 74, with a median age of 59. The cohort was predominantly female (69%), with males making up 31%. Half of the participants (50%) were classified as WHO Grade 1, and 19% had no available histological data. Regarding treatment, 81% received normal fractionation therapy with doses of 1.8 Gy. Over 90% of patients found the information clear, accurate, and consistent with their experiences, with 65% believing it would have been helpful prior to treatment. Additionally, 60% trusted the information and expressed interest in using ChatGPT-4 for future medical inquiries. Clinicians rated the relevance and accuracy of most questions with an average Likert score above 4, but the completeness was rated slightly lower (below 4), especially for questions regarding specific therapy details and side effects. Conclusion: ChatGPT-4 demonstrates potential as a supplementary educational tool for meningioma patients, though certain areas may require improvement. Patients reported the information as clear and relevant, while clinicians identified gaps in the completeness of treatment-specific details. With more than half of patients trusting the information and showing interest in using LLMs for future medical inquiries, these findings advocate for a discussion on integrating LLMs into patient education. However, their use should be complemented by professional guidance to ensure accuracy and comprehensiveness.

Keywords: meningioma,large language model,radiation therapy