S1479
Interdisciplinary - Patient involvement
ESTRO 2026
Munich, Germany. 4 Institute of Radiation Medicine, Helmholtzzentrum München, Munich, Germany
Purpose/Objective: As structured decision-aids for patients facing radiotherapy (RT) are largely missing, many patients use non-validated internet sources. In the current project funded by the Bavarian Center for Cancer Research (BZKF) we developed and validated a Large Language Model (LLM)-enabled artificial intelligence (AI) system, designed to provide better information on RT for breast cancer patients. Material/Methods: An LLM-enabled AI system was developed jointly by radiation oncology experts and technical AI specialists, by generating curated content based on breast cancer guidelines, studies and medical manuals. Information was consolidated into a single document in a question-and-answer format and presented in language accessible to patients providing clear, validated, and trustworthy information.Usability testing was performed with 30 participants (5 radiation oncologists, 5 physicians, 5 medical staff, 5 medical students and 10 laypersons). Results: Participants rated answer quality of five LLM- generated answers to therapy-related questions, regarding trustworthiness, comprehensibility, and scope using a 6-point Likert scale (1–6, 1=very good, 6=very bad). Average ratings were transformed into percentages using a linear conversion, with 1 mapped to 100% and 6 mapped to 0%. Results across all dimensions were satisfactory: trustworthiness 95%, comprehensibility 93%, and scope 87%. In the target group of patient representatives, the overall rating was 93.6% (Figure 1). Participants also evaluated a validated 5-point Likert-scale questionnaire on usability and other chatbot related questions (Figure 2). Free-text feedback confirmed the clarity and accessibility of information, while also highlighting challenges in balancing medical precision with patient- friendly language.
References: Victorson D, et al. Int J Radiat Oncol Biol Phys. 2024;129(3):684–693.Trommer M, et al. Cochrane Database Syst Rev. 2023;(3):CD013448.Lawen T, et al. Cancers (Basel). 2024;16(5):958.Bailey C, et al. Int J Environ Res Public Health. 2022;19(10):5373.Ilie G, et al. J Cancer Surviv. 2023;17(2):393–404. Keywords: mindfulness, prostate radiotherapy, PROM Poster Discussion 1952 Development and Validation of an LLM-based decision Aid providing Evidence-Based Information on Radiotherapy for Breast Cancer Patients Irina S. Graf 1 , Sophia Kiesl 1 , Max Tschochohei 1,2 , Luisa Allwohn 1 , Sophie T. Behzadi 1 , Jacqueline Lammert 3 , Sophie Maier 1 , Rebecca Moser 1 , Jana Nano 1 , Stephanie E. Combs 1,4 , Kai J. Borm 1 1 Department of Radiation Oncology, Klinikum Rechts der Isar, Technial University of Munich (TUM), Munich, Germany. 2 Google Cloud, Google, Munich, Germany. 3 Department of Gynecology and Obstetrics, Klinikum Rechts der Isar, Technial University of Munich (TUM),
Made with FlippingBook - Share PDF online