ConoCancer: A RAG-Based Chatbot for Breast Cancer Health Literacy
Abstract
Breast cancer remains one of the most commonly diagnosed cancers among women worldwide, yet gaps in health literacy and language access continue to limit early detection and informed decision-making, particularly among low-literacy Hispanic/Latina populations. Conversational AI offers on-demand health education at scale, but general-purpose large language models deployed without grounding frequently produce fluent yet unsupported or outdated responses, a failure that is especially costly in medical contexts. A further complication is that many patients in this population communicate by code-switching between English and Spanish rather than in one language exclusively. To address these gaps, we developed a bilingual intelligent tutoring system for breast cancer education, constituting short instructional videos with interactive quizzes and a chatbot for answering foundational breast cancer questions. However, this had some limitations that are targeted in this study. This paper focuses on improving the Quiz Scoring module and the Chatbot. We replaced the existing chatbot's architecture with a Retrieval-Augmented Generation (RAG) pipeline and keyword-based quiz grading with semantic grading and natural language inference (NLI). We evaluate the performance across four language variants (English, Spanish, and code-switched variants). Generation quality is assessed with Ragas (RAG Assessment), and retrieval performance is measured against the source passage from which questions were drawn. Generation was found comparable across language variants, but retrieval was not. The Spanish and code-switched queries could retrieve the corresponding passage, but failed to rank it as first option. Semantic grading outperformed the keyword baseline for only English, while NLI outperformed both approaches for all variants.
Keywords: Retrieval-Augmented Generation, Breast Cancer Education, Health Literacy, Code-Switching
DOI: 10.54941/ahfe1008437
Cite this paper
More from this volume
- AI-Enhanced Live-Streamed English Language Learning: Designing and Evaluating an Intelligent Platform for Chinese Learners
- "They Will Move": Confirmation Bias, Expectation Frames, and Premature Monitoring Drop Off in COLREG Encounters
- Automation transparency communication does not improve supervisory control performance, but shapes operator trust
- The influence of neuroticism on startle and surprise response: a flight simulator study
- EEG-Based Classification of Reward, Penalty, and Movement Events Across Four Video Games of Varying Complexity
- The Effects of Task Similarity and Time Pressure on Task Performance Following Interruptions in Complex Human-Computer Interaction Systems
- Trusting the Chain, Not the Black Box: A Provenance-Based Architecture for Human-Machine Teaming in Military Autonomy
- Design of a Simplified Humanoid Robot Face: Cultural Differences and Emotional Synchronization
- From Visual Observation to Auditory Interaction: A Human-Centered Tangible Music System Based on Physical Objects
- The Impact of Scarcity-Based Dark Patterns on User Performance, Perceived Darkness, and Workload in Subscription Services
- A Behavioral Observation Tool for Diagnosing Human–AI Interaction Profiles in Engineering Design Education
- The Mediated Moon: Semantic Mediation Between Gesture and Generative Image


AHFE Open Access