A Retrieval-Augmented Generation Framework for Conversational Chronic Disease Monitoring Agents in Resource-Constrained Environments

This paper introduces Chrono, a conversational agent powered by Retrieval- Augmented Generation (RAG) for proactive monitoring of chronic conditions like diabetes and heart disease. The system analyzes patient data to detect risks and deliver personalized recommendations. Chrono integrates a local knowledge base with large language models, incorporating rule-based alerts and multi-case conversational memory for context-aware interactions. A comparative evaluation, validated by a certified medical professional in a blind review, found Chrono’s recommendations to be 15-18% more clinically appropriate and actionable than a baseline rule-based system, highlighting its potential for scalable deployment.