Main Session
Sep
29
PQA 07 - Head and Neck Cancer, Lung Cancer/Thoracic Malignancies, and Nursing and Supportive Care
3512 - AI-Driven Nutrition Support Demonstrates High Usability and Cultural Relevance Among Radiation Therapy Cancer Survivors: Early Results from the CANEAT-IMPACT Study
Presenter(s)
Adam Katzenberg, DO, MBA - Thomas Jefferson University, Philadelphia, PA
A. M. Katzenberg, A. M. Allimulla, M. Thapar, N. Francois, J. Jacoby, J. Logan, J. Joshi, M. Gupta, J. Mallon, R. Simon, D. Thomas, W. Choi, and N. L. Simone; Dept. of Radiation Oncology, Sidney Kimmel Medical College and Comprehensive Cancer Center, Thomas Jefferson University, Philadelphia, PA
Purpose/Objective(s):
Optimal nutrition is critical for improving cancer outcomes and quality of life in many cancer survivors. However, access to personalized nutritional counseling is limited. We hypothesized that the CANEAT application (app), a novel large language model (LLM)- based tool developed by our team, would provide usable, culturally relevant, and affordable dietary recommendations to support radiation therapy (RT) cancer survivors in managing long-term nutritional needs.Materials/Methods:
An IRB-approved prospective qualitative study was designed to enroll cancer survivors who had previously received RT to participate in focus groups to test a novel app developed by radiation oncologists, nutritional scientists, and physicists. Participants engaged in a single-session focus group that consisted of a 1-on-1 guided interaction with study staff and the survivor to use the CANEAT app. Through interaction with the app, participants input demographic, geographic, and cultural information. The LLM then generates personalized meal plans and grocery lists tailored to cultural preferences, budgets, and local access. Patients evaluated their meal plans, then completed the validated System Usability Scale (SUS) and structured focus group questions. The primary endpoint was a SUS usability score >70, indicating acceptability. Group-structured questions yielded qualitative feedback on relevance, limitations, and intent to use. Thematic analysis was applied to focus group transcripts. Data were analyzed using descriptive statistics for SUS scores and NVivo for qualitative themes.Results:
There were 8 total enrolled survivors (62.5% were 65 years or older; 100% were female; 62.5% were racial/ethnic minorities; 100% with prior RT), all completed the session. The mean SUS score was 82.5 (SD 14.3; range 60-100), with 85% rating the app as acceptable (>70). 87.5% agreed that the app helped them make healthier choices, simplified meal planning, and offered meal plans that could be realistically followed during treatment. Qualitative themes highlighted that the app provided meal plans with high cultural relevance, the ability to have grocery lists within budget constraints, and geographic feasibility. The limitation was in addressing RT side effects.Conclusion:
The CANEAT app demonstrated strong usability and represents a scalable, AI-driven approach to addressing nutritional disparities in cancer survivorship care. These early findings support evaluation in larger, diverse cohorts and highlight the potential of LLM-based tools to deliver equitable, personalized nutrition support that may improve quality of life for cancer survivors.