Scholarlink

The proposed multimodal medical chatbot combines NLP and image recognition to offer timely healthcare consultations. Through text and image inputs, it aims to diagnose diseases and provide basic information ahead of doctor consultations. Utilizing multimodal RAG enhances its responsiveness by retrieving relevant information from uploaded documents. Integration with PubMed Search allows users to access the latest medical research, aiding queries on new treatments or breakthroughs. Real-Time Web scraping supplements its knowledge base by gathering information from relevant webpages. The chatbot's objective is to reduce healthcare costs and improve medical knowledge accessibility. By furnishing preliminary information, it empowers users to make informed health decisions before seeking professional advice. The performance of the chatbot model was evaluated through human-based validation where appraisers rate its responses based on relevance, coherence, and no-hallucination. The combined module achieves impressive scores in human-based validation: 0.9 for relevance, 0.8 for coherence, and 0.9 for no-hallucination, demonstrating its proficiency in generating highly relevant, coherent, and non-hallucinatory responses.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced Medical Multimodal Retrieval Augmented Generation (RAG) Application

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • J. Anitha,
  • R. Avanthika,
  • B. Kavipriya,
  • S. Vishnupriya

摘要

The proposed multimodal medical chatbot combines NLP and image recognition to offer timely healthcare consultations. Through text and image inputs, it aims to diagnose diseases and provide basic information ahead of doctor consultations. Utilizing multimodal RAG enhances its responsiveness by retrieving relevant information from uploaded documents. Integration with PubMed Search allows users to access the latest medical research, aiding queries on new treatments or breakthroughs. Real-Time Web scraping supplements its knowledge base by gathering information from relevant webpages. The chatbot's objective is to reduce healthcare costs and improve medical knowledge accessibility. By furnishing preliminary information, it empowers users to make informed health decisions before seeking professional advice. The performance of the chatbot model was evaluated through human-based validation where appraisers rate its responses based on relevance, coherence, and no-hallucination. The combined module achieves impressive scores in human-based validation: 0.9 for relevance, 0.8 for coherence, and 0.9 for no-hallucination, demonstrating its proficiency in generating highly relevant, coherent, and non-hallucinatory responses.