SECURE AND EXPLAINABLE RETRIEVAL-AUGMENTED GENERATION FOR HEALTHCARE DECISION SUPPORT

Abstract

Healthcare organizations increasingly rely on Artificial Intelligence (AI) to support clinical decision-making, healthcare claims processing, prior authorization, policy interpretation, and patient care coordination. However, conventional Large Language Models (LLMs) often suffer from hallucinations, limited explainability, outdated knowledge, and security concerns when deployed in healthcare environments. Retrieval-Augmented Generation (RAG) addresses these limitations by combining enterprise knowledge retrieval with generative AI to produce context-aware and evidence-supported responses. This paper proposes a secure and explainable RetrievalAugmented Generation framework for healthcare decision support by integrating enterprise healthcare knowledge repositories, vector databases, explainable AI, secure cloudnative infrastructure, machine learning, and intelligent retrieval mechanisms. The proposed framework continuously retrieves authoritative clinical evidence, healthcare policies, reimbursement guidelines, and regulatory documentation before generating transparent healthcare recommendations. Experimental analysis demonstrates significant improvements in retrieval precision, decision accuracy, explainability, regulatory compliance, operational efficiency, and enterprise scalability compared with conventional healthcare AI systems. Keywords: Retrieval-Augmented Generation, Explainable AI, Healthcare Decision Support, Large Language Models, Healthcare Security, Vector Database, Enterprise AI, Semantic Search, Machine Learning, Cloud-Native Computing.