SECURE RETRIEVAL-AUGMENTED GENERATION FOR PRIVACY-AWARE HEALTHCARE CLAIMS INTELLIGENCE

Abstract

Healthcare claims processing requires continuous access to sensitive patient information, payer policies, clinical documentation, and regulatory guidelines. While Retrieval-Augmented Generation (RAG) significantly improves the accuracy of generative artificial intelligence by grounding responses in enterprise knowledge, healthcare environments demand strong privacy protection, secure data access, and regulatory compliance throughout the retrieval and generation process. This paper proposes a Secure Retrieval-Augmented Generation framework integrating semantic retrieval, vector databases, privacy-preserving knowledge management, enterprise security controls, automated governance, and cloud-native MLOps for healthcare claims intelligence. The proposed architecture combines secure document ingestion, encrypted vector indexing, role-based access control, evidence-grounded response generation, and continuous security monitoring to support intelligent healthcare claims decisionmaking. Experimental evaluation demonstrates improvements in retrieval precision, response accuracy, security compliance, privacy preservation, operational scalability, and administrative efficiency. The proposed framework provides a secure, explainable, and production-ready solution for privacy-aware healthcare claims intelligence.