Please use this identifier to cite or link to this item: https://www.um.edu.mt/library/oar/handle/123456789/145966
Title: AI-driven chatbot for mobile-based supportive care in breast cancer : enhancing patient treatment outcomes through digital intervention
Authors: Sammut, Maria Stella (2026)
Keywords: Breast -- Cancer -- Patients -- Malta
Breast -- Cancer -- Radiotherapy
Breast -- Cancer -- Patients -- Services for
Artificial intelligence -- Medical applications
Chatbots -- Malta
Natural language processing (Computer science)
Issue Date: 2026
Citation: Sammut, M. S. (2026). AI-driven chatbot for mobile-based supportive care in breast cancer : enhancing patient treatment outcomes through digital intervention (Master’s dissertation).
Abstract: This dissertation investigates the potential of RadPal, an AI-powered chatbot developed to support breast cancer patients undergoing radiation therapy in Malta. Radiation therapy is a key treatment for breast cancer, improving local control while often causing side effects that diminish patients' quality of life. Many patients face challenges accessing adequate supportive care for their physical and psychosocial needs. The study aimed to assess RadPal's efficacy in meeting patients' information needs. Built on the DiFy platform using retrieval-augmented generation (RAG) techniques, ChatGPT-4.0, and data from reputable sources, local hospital services, and a fictitious database, RadPal delivers personalized, empathetic responses to queries on treatment, side effects, and appointments. Evaluation involved ten healthcare professionals (HCPs) who assessed the prototype via an online questionnaire from the Society of Radiographers Malta. It included task-level usefulness ratings (5-point Likert scale), the System Usability Scale (SUS), Chatbot Usability Questionnaire (CUQ), and open-ended feedback. Data were analysed descriptively and thematically. Results showed strong usability (mean SUS: 82; mean CUQ: 75.9), with high ratings for side-effect management (4.4/5), treatment information (4.3/5), and emotional support (4.3/5). Patient-specific information scored lower (3.2/5), due to limited training data. HCPs praised ease of use (n=5) and empathy (n=3), but suggested improvements in response speed (n=4), multilingual support (n=2), and localized service details (n=3). Overall, RadPal demonstrates high potential to empower patients, reduce underreporting of side effects, and ease healthcare burdens, with possible annual savings. Future developments should integrate electronic health records, enhance accessibility, and conduct larger studies to validate the accuracy of the responses and assess the real-world impact on patient care following the introduction of these initiatives.
Description: M.Sc.(Melit.)
URI: https://www.um.edu.mt/library/oar/handle/123456789/145966
Appears in Collections:Dissertations - FacICT - 2026

Files in This Item:
File Description SizeFormat 
2618ICTIFC500105069362_1.PDF
  Restricted Access
2.94 MBAdobe PDFView/Open Request a copy


Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.