AM-RAF: An Agentic, Retrieval-Augmented, Uncertainty-Gated Framework for Trustworthy Clinical Decision Support in Resource-Constrained Health Systems
Abstract
Abstract Large language models (LLMs) now approach or exceed expert-level performance on standardised medical examinations, and retrieval-augmented generation (RAG) substantially reduces their tendency to fabricate clinical facts. Independently, agentic architectures that interleave reasoning with tool use have begun to appear in medicine as multi-agent diagnostic assistants. Yet these three advances — grounded retrieval, agentic reasoning, and calibrated uncertainty have not been composed into a single, formally specified architecture aimed at the specific constraints of point-of-care diagnosis in low- and middle-income countries (LMICs), where connectivity is intermittent, clinician-to-patient ratios are low, and demographic representation in training data is often poor. This paper proposes the Agentic Multimodal Retrieval-Augmented Framework (AM-RAF), a conceptual six-layer architecture that grounds a cooperating set of diagnostic agents in a locally curated clinical knowledge base, gates every output through a hallucination- and confidence-aware deferral mechanism before it reaches a patient, and periodically recalibrates itself through privacy-preserving federated feedback rather than centralising patient data. We formalise the diagnostic-assistance problem as an acuity-weighted expected cost minimisation between autonomous answering and clinician deferral, specify the six architectural layers and their interfaces, and present a closed-loop algorithm for online…
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.