Abstract: FR-PO0031
Developing an Artificial Intelligence (AI) Chatbot That Assists Patients in Finding Transplantation Centers That Provide Transplants for Patients Like Them
Session Information
- Artificial Intelligence and Digital Health at the Bedside
November 07, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Lozano, Cinthia, University of Texas Medical Branch, Galveston, Texas, United States
- Jagannatha Kagathi, Jithendra, University of Minnesota, Minneapolis, Minnesota, United States
- McKinney, Warren T., Hennepin Healthcare Research Institute, Minneapolis, Minnesota, United States
- Anchondo Ley, Diego Mauricio, University of Texas Medical Branch, Galveston, Texas, United States
- Vo, Duy, University of Texas Medical Branch, Galveston, Texas, United States
- John, Gladson S, University of Texas Medical Branch, Galveston, Texas, United States
- Pakhomov, Serguei, University of Minnesota, Minneapolis, Minnesota, United States
- Schladt, David P., Hennepin Healthcare Research Institute, Minneapolis, Minnesota, United States
- Snyder, Jon J., Hennepin Healthcare Research Institute, Minneapolis, Minnesota, United States
- Matas, Arthur J., University of Minnesota, Minneapolis, Minnesota, United States
- Israni, Ajay K., University of Texas Medical Branch, Galveston, Texas, United States
Background
The website transplant center search (TCS) provides patients with personalized data from the Scientific Registry of Transplant Recipients (SRTR) to help them identify and compare transplant centers. Previous research has shown that some patients may need assistance navigating the TCS. Therefore, we are developing a HIPAA-compliant chatbot to improve navigation and further assist patients in making informed decisions about transplant centers.
Methods
We are implementing a multimodal, bilingual (English and Spanish) chatbot using the Llama 3.1 70B large language model hosted on a secure institutional compute platform and the Dialog Tree dialogue management framework to mitigate limitations of the model. The chatbot underwent continuous model performance evaluation during development.
Results
The chatbot guides patients to navigate the TCS website. The chatbot asks some medical questions (Figure 1) and shows a personalized search result using data from the SRTR (Figure 2). It provides information regarding their options in accordance with their medical condition.
Conclusion
The TCS is the first chatbot using an LLM to provide personalized search results to patients and their family members across the U.S. This patient-centered approach aims to improve access to transplantation and in the future could help patients make use of nationwide data from the SRTR.
Funding
- Other NIH Support