AI-CChaSE: AI-enabled Cancer CHatbots for Symptom Education

/ABSTRACT This proposal addresses an important, unmet need in cancer symptom management: cancer patients are routinely impacted by chronic symptoms caused by their disease and also by cancer-directed treatment. These long-term sequalae of cancer and cancer-therapy can be devastating to patients' physical, psychological, social, and financial wellbeing. However, high-quality symptom care is often limited by patient accessibility, patient and clinician education, and communication with and among the care team. Chatbots for symptom monitoring and management have been shown to improve patient empowerment, healthcare utilization, and outcomes, but have previously been limited to rules-based systems which can be inflexible, require significant manual effort to develop, and cannot easily adapt to new guidelines and user preferences. The overarching objective of this proposal will be to advance new methods and benchmarks for the development of guideline-grounded chatbots for improved cancer symptom education and communication. Our central innovation will be the research and development of efficient, extensible, and user-centered methods that leverage advances in large language models (LLMs) to convert clinical resources into usable, interactive technologies with improved performance and accessibility. In Specific Aim 1, we will improve the factuality of LLM question-answering about symptoms, which we will approach by developing new methods to link LLMs with knowledge in multi-modal clinical practice guidelines. In Specific Aim 2, we will investigate high-fidelity methods for LLM-based text simplification of recommendations in guidelines and the electronic health records to accommodate the different needs of end- users. In Specific Aim 3, we will study the needs, perceptions, concerns, and ethical considerations of patients and clinicians, which will inform technical developments is Specific Aims 1/2 and guide user-centered design for chatbot interfaces. The resulting interface will be evaluated in empirical studies. Patient and clinician stakeholders will be centered throughout all aims of our research. This work will be highly significant and innovative because it uses advances in artificial intelligence to amplify the availability of reliable information resources for cancer symptom care. These methods may thereby improve cancer outcomes and quality-of-life, while providing broad generalizable insights for the use of artificial intelligence-based information and education resources across biomedical fields. Project Number: 1R01CA311924-01 | Fiscal Year: 2026 | NIH Institute/Center: National Cancer Institute (NCI) | Principal Investigator: Danielle Bitterman (+2 co-PIs) | Institution: BRIGHAM AND WOMEN'S HOSPITAL, BOSTON, MA | Award Amount: $819,628 | Activity Code: R01 | Study Section: Clinical Informatics and Digital Health Study Section[CIDH] View on NIH RePORTER: https://reporter.nih.gov/project-details/11344048