Conversational Health Agents: A Personalized Large Language Model-Powered Framework

This project develops openCHA, an open-source framework for building next-generation conversational health agents (CHAs). Unlike traditional chatbots, openCHA enables multistep reasoning, multimodal integration, personalized conversations, and explainability. Researchers and developers can use the framework to create domain-specific CHAs that interact with diverse AI platforms and external health data. Demonstrated use cases include diabetes management (92.1% accuracy vs. GPT-4’s 51.8%), food recommendations, mental health chatbot evaluation, empathy-based emotion recognition (89% accuracy), and physiological data analysis of PPG signals. By advancing personalization, reliability, and scalability, openCHA empowers the development of trustworthy, adaptable CHAs that support a wide range of healthcare tasks.

Publications:


Full paper

Project information

  • Category: Future Healthcare, Large Language Models, and Personal Health Models
  • Contact Person: Ramesh Jain
  • Conversational Health Agents: A Personalized Large Language Model-Powered Framework

info@futurehealth.uci.edu

© Copyright 2021 UCI Institute for Future Health - All Rights Reserved