Open Access

KG4Diagnosis: A Hierarchical Multi-Agent LLM Framework with Knowledge Graph Enhancement for Medical Diagnosis

1 University of Warwick
2 Cranfield University
3 University of Cambridge

Abstract

Integrating Large Language Models (LLMs) in healthcare diagnosis demands systematic frameworks that can handle complex medical scenarios while maintaining specialized expertise. We present KG4Diagnosis, a novel hierarchical multi-agent framework that combines LLMs with automated knowledge graph construction, encompassing 362 common diseases across medical specialties. Our framework mirrors real-world medical systems through a two-tier architecture: a general practitioner (GP) agent for initial assessment and triage, coordinating with specialized agents for in-depth di-agnosis in specific domains. The core innovation lies in our end-to-end knowledge graph generation methodology, incor-porating: (1) semantic-driven entity and relation extraction optimized for medical terminology, (2) multi-dimensional de-cision relationship reconstruction from unstructured medical texts, and (3) human-guided reasoning for knowledge ex-pansion. KG4Diagnosis serves as an extensible foundation for specialized medical diagnosis systems, with capabilities to incorporate new diseases and medical knowledge. The framework’s modular design enables seamless integration of domain-specific enhancements, making it valuable for devel-oping targeted medical diagnosis systems. We provide archi-tectural guidelines and protocols to facilitate adoption across medical contexts.

How to Cite

Zuo, K., Jiang, Y., & Mo, F. (2026). KG4Diagnosis: A Hierarchical Multi-Agent LLM Framework with Knowledge Graph Enhancement for Medical Diagnosis. Asia Journal of Social Innovation and Development, 2(1), 10. Retrieved from https://www.ajsid.org/index.php/pub/article/view/30

References

📄 Abdulla, K.; Mukherjee, S.; and Ranganathan, P. 2023. Inte-grating Multimodal Data for Enhancing Knowledge Graphs:Current Challenges and Opportunities. Journal of Big Data,10(1): 1–15.
📄 Al Khatib, H. S.; Neupane, S.; Kumar Manchukonda, H.;Golilarz, N. A.; Mittal, S.; Amirlatifi, A.; and Rahimi, S.2024. Patient-centric Knowledge Graphs: A Survey of Cur-rent Methods, Challenges, and Applications. Frontiers in Artificial Intelligence, 7: 1388479.
📄 Alam, F.; Giglou, H. B.; and Malik, K. M. 2023. Auto-mated Clinical Knowledge Graph Generation Framework for Evidence-based Medicine. Expert Systems with Appli-cations, 233: 120964.
📄 Alsentzer, E.; Murphy, J.; Boag, W.; Weng, W.; Jin, D.; Nau-mann, T.; and McDermott, M. 2019. Publicly Available Clinical BERT Embeddings. arXiv:1901.08746.
📄 Amos, L.; Anderson, D.; Brody, S.; Ripple, A.; and Humphreys, B. L. 2020. UMLS Users and Uses: A Current Overview. Journal of the American Medical Informatics As-sociation, 27(10): 1606–1611.