PRODUCTION-READY RAG ARCHITECTURE FOR EVIDENCE-GROUNDED HEALTHCARE ADMINISTRATIVE AUTOMATION

Abstract

Healthcare organizations continuously process large volumes of administrative data related to patient registration, insurance verification, claims processing, prior authorization, clinical documentation, billing, and regulatory compliance. Traditional healthcare administrative systems depend heavily on manual workflows, rule-based automation, and isolated information systems, resulting in delayed processing, inconsistent decision-making, increased operational costs, and administrative burden. Recent advancements in RetrievalAugmented Generation (RAG) provide an effective approach for combining large language models with trusted healthcare knowledge repositories to generate evidence-grounded and explainable administrative decisions. This paper proposes a production-ready RetrievalAugmented Generation architecture for intelligent healthcare administrative automation. The proposed framework integrates electronic health records, insurance policies, medical coding standards, clinical practice guidelines, regulatory documents, and enterprise knowledge repositories with advanced retrieval mechanisms and large language models. The architecture supports accurate document understanding, intelligent workflow automation, evidence-based decision support, and continuous knowledge updating. Experimental analysis demonstrates improved administrative efficiency, decision consistency, explainability, regulatory compliance, and operational scalability for modern healthcare organizations. Keywords: Retrieval-Augmented Generation, Healthcare Administration, Large Language Models, Electronic Health Records, EvidenceBased Decision Support, Medical Coding, Artificial Intelligence, Healthcare Automation.