EXPLAINABLE RAG SYSTEMS FOR TRANSPARENT HEALTHCARE CLAIMS REVIEW AND DECISION SUPPORT

Abstract

Healthcare claims review requires accurate interpretation of payer policies, clinical documentation, coding standards, and regulatory guidelines while maintaining transparency and accountability in administrative decisionmaking. Conventional rule-based systems often lack adaptability, whereas standalone generative artificial intelligence models may produce nontransparent recommendations unsupported by verifiable evidence. Explainable RetrievalAugmented Generation (RAG) combines semantic knowledge retrieval with explainable artificial intelligence to generate evidencegrounded recommendations supported by authoritative enterprise documents. This paper proposes an Explainable RAG framework integrating semantic retrieval, vector databases, enterprise knowledge management, explainability modules, cloud-native MLOps, and automated governance for transparent healthcare claims review and decision support. The proposed architecture provides traceable reasoning, confidence estimation, evidence citation, and policy-based explanations throughout the claims review process. Experimental evaluation demonstrates improvements in retrieval precision, decision transparency, explainability, operational efficiency, regulatory compliance, and administrative productivity. The proposed framework offers a scalable, trustworthy, and production-ready solution for intelligent healthcare claims decision support. Keywords— Explainable AI, RetrievalAugmented Generation, Healthcare Claims, Decision Support, Semantic Retrieval, Large Language Models, MLOps, Healthcare Policy