A Hybrid Retrieval-Augmented Generation Framework for Intelligent Symptom Monitoring and Disease Guidance

Abstract

Early identification of diseases based on patient-reported symptoms can improve timely medical guidance and reduce the burden on healthcare services. This paper presents SymptoTrackAI, an intelligent symptom monitoring framework that integrates Retrieval-Augmented Generation (RAG) with FAISS-based similarity search to provide context-aware disease suggestions from user-entered symptoms. The proposed system utilizes a symptom–disease knowledge base and employs the Sequence-NQ transformer to retrieve semantically relevant medical information before generating an appropriate response. A secure web application is developed with modules for user registration, authentication, AI-assisted symptom analysis, and historical symptom log management. User interactions and predicted outcomes are stored to support continuous health monitoring and future analysis. The retrieval mechanism enables the chatbot to identify diseases by matching symptom descriptions with medically relevant records, thereby improving the relevance of generated recommendations. Experimental observations demonstrate that the hybrid retrieval framework consistently delivers accurate and context-sensitive responses for diverse symptom queries while maintaining efficient response time. The proposed solution offers a practical, scalable, and userfriendly approach for preliminary symptom assessment, enabling individuals to obtain timely health-related guidance and maintain a digital record of symptom history for informed healthcare decision-making.