S2839
RTT - RTT education, training, and advanced practice
ESTRO 2026
Clear, consistent communication of information is central to patient centred radiotherapy, yet patients often struggle to understand the information provided about treatment. Generative pre-trained transformers (GPT) models may offer a solution to enhance patient communication and support safe, innovation led practice. This study aims to develop a generative artificial intelligent (AI) chatbot as an innovative patient information resource and to evaluate its
with the quality of RTT-generated contours. Most frequent challenges reported by RTTs included anatomical variability, image artefacts, and a lack of a standardised contouring protocol. The seminal vesicles and bladder neck were identified as the most challenging structures to delineate. Despite these challenges, RTTs achieved a mean contouring time of 13.45 min [Fig. 1], consistently demonstrating reliability.
feasibility and safety. Material/Methods:
We developed RAYA (Radiotherapy and You Assistant) on the ChatGPT platform as a customised, trained AI model grounded in institutionally and nationally approved radiotherapy information resources (Figure 1). We constrained RAYA outputs to verified, context- specific knowledge enhanced accuracy and relevance. A multidisciplinary team led by a Radiation Therapist carefully curated the knowledge base, authored system instructions, and embedded safety parameters to ensure responses were source-cited, avoided diagnostic or prescribing language, included non- clinical disclaimers, and escalated risk scenarios.
Fig.1 - RTT contouring time per patient (n=31) per fraction (n=142) Conclusion: An RTT-led MRgART workflow for prostate cancer was successfully implemented, enhancing operational efficiency and opening the path for advanced practice among RTTs. This model supports a clinician-lite workflow, fostering RTTs professional development while maintaining high standards of care. Throughout the implementation, treatment time slots were safely reduced from 60 min to 45 min, with potential for further optimization. This offers a scalable roadmap for other tumor sites and CT-based online adaptive radiotherapy. Keywords: MRI-Guided, RTT-led, Prostate Development and Evaluation of a Generative AI Information Chatbot for Patients Undergoing Radiation Therapy Sany Dangol 1 , Sinead Brennan 1 , Gerard G Hanna 1,2 , Roisin O Maolalai 1,3 , Ciaran Malone 1,4 1 Radiation Oncology, St. Luke’s Radiation Oncology Network, Dublin, Ireland. 2 2. Trinity St. James’s Cancer Institute, Trinity College Dublin, Dublin, Ireland. 3 3. Irish Research Radiation Oncology Group, 3. Irish Research Radiation Oncology Group, Dublin, Ireland. 4 4. Erasmus MC Cancer Institute, University Medical Center Rotterdam, Department of Radiotherapy, Dublin, Ireland Proffered Paper 4317 Empathy Meets Innovation: An RTT-led
Multidisciplinary radiotherapy staff and patient/public representatives were recruited through institutional and patient and public involvement (PPI) networks. Each participant interacted with RAYA and completed an online questionnaire recording three patient style prompt-answer pairs and a combination of yes/no and Likert ratings for assessing domains such as usability, accuracy, communication quality, safety/escalation and boundaries. Quantitative data were analysed descriptively, and qualitative comments from open text feedback were grouped thematically. Results: Eighteen participants completed the questionnaire, staff (n=10) and patient/public (n=8) participants. Quantitative findings demonstrated consistently positive perceptions across all domains. RAYA was easy to access, with smooth interaction flow and excellent usability (median=5, IQR=4-5). Communication quality and empathy were rated as clear (median=5, IQR=4-5) and respectful (median=5, IQR=5-5). Safety and accuracy of responses were consistently high (median=5, IQR=4-5), with 100% correct referencing and no hallucinations or unsafe content detected. RAYA escalated and stated
Purpose/Objective:
Made with FlippingBook - Share PDF online