QuickAid: A Hybrid RAG and LLM Framework for Improving Chatbot Accuracy and Relevance in First-Aid Guidance

India is facing significant healthcare issues, including a doctor-patient ratio of 1:836, delayed emergency response times, and terribly limited medical infrastructure in rural areas where 65% of the population reside, while healthcare facilities number only 25%. With overcrowded hospitals dealing with routine non-emergency cases, proper first aid counsel could help in such cases. Also, in case of emergencies step by step first aid guidance help patients to save their life by managing critical situations, like cardiac arrest, choking, or severe bleeding before reaching hospitals. To bridge the healthcare gap, there is a need of medical chatbot that provides immediate, cost effective and multimodal emergency guidance especially in rural areas. The rapid advancements in artificial intelligence(AI) have significantly transformed healthcare applications, particularly in the development of medical chatbots. To create an intelligent First Aid Medical Chatbot, this study proposes QuickAid, a hybrid approach that integrates Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs). RAG improves the ability of the chatbot to provide accurate, real-time first aid advice by retrieving relevant medical knowledge from authoritative and evidence-based sources and combining it with generative AI capabilities. This ensures that users receive accurate, contextually relevant, and timely emergency support. The dataset is designed to cover essential first aid requirements, including choking, headaches, and other medical emergencies. The chatbot facilitates multimodal interaction through both text and audio for better accessibility. Its hybrid strategy helps to lower the risk of errors by employing structured retrieval. QuickAid's effectiveness is supported by a BLEU score of 0.87 and ROUGE scores of 0.92 (ROUGE-1), 0.85 (ROUGE-2), and 0.91 (ROUGE-L), reflecting its high accuracy and fluency.