From Tables to Graphs with ClinicoAtlas: Leveraging LLMs to Support Modeling and Mining Knowledge Graphs

• Rafael C. G. Conrado University of São Paulo (USP)
• Eduardo M. Campos University of São Paulo (USP)
• Caetano Traina Jr. University of São Paulo (USP) https://orcid.org/0000-0002-6625-6047
• Agma J. M. Traina University of São Paulo (USP) https://orcid.org/0000-0003-4929-7258
• Mirela T. Cazzolato University of São Paulo (USP) https://orcid.org/0000-0002-4364-010X

DOI: https://doi.org/10.5753/sbbd_estendido.2025.247610

Abstract

Given a set of Electronic Health Records organized as tabular data, how can we model and evaluate relationships between various concepts related to health conditions and treatments? Representing EHRs as Graphs allows a much more intuitive interpretation of relationships between concepts and permits the mining of node features, generating relevant metrics for hospital data analysis. However, these metrics are often generic and difficult to interpret. In this context, Large Language Models can assist in constructing these graphs by defining the semantic relationships between concept pairs and attributing contextual meaning to the extracted features. The proposed ClinicoAtlas tool integrates LLMs into the construction of knowledge graphs, facilitating the extraction and interpretation of meaningful information for healthcare management.

Keywords: Knowledge graphs, LLMs, electronic health records, graph features, medical data

References

Clusmann, J., Kolbinger, F., Muti, H., Carrero, Z., Eckardt, J.-N., Laleh, N., Löffler, C., Schwarzkopf, S.-C., Unger, M., Veldhuizen, G., Wagner, S., and Kather, J. (2023). The future landscape of large language models in medicine. Communications medicine, 3:141. DOI: 10.1038/s43856-023-00370-1.

Conrado, R., Gutierrez, M., Jr., C. T., Traina, A., and Cazzolato, M. (2024). Combining semantic graph features and a common data model to exploit the interoperability of patient databases. In Anais do XXXIX Simpósio Brasileiro de Bancos de Dados, pages 701–707, Porto Alegre, RS, Brasil. SBC. DOI: 10.5753/sbbd.2024.243153.

de Lima, D. M. et al. (2019). Transforming two decades of ePR data to OMOP CDM for clinical research. In MEDINFO 2019, volume 264, pages 233–237. IOS Press. DOI: 10.3233/SHTI190218.

Meyer, L.-P., Stadler, C., Frey, J., Radtke, N., Junghanns, K., Meissner, R., Dziwis, G., Bulert, K., and Martin, M. (2024). LLM-assisted Knowledge Graph Engineering: Experiments with ChatGPT. Springer Fachmedien Wiesbaden. DOI: 10.1007/978-3-658-43705-3_8.

Murali, L., Gopakumar, G., Viswanathan, D. M., and Nedungadi, P. (2023). Towards electronic health record-based medical knowledge graph construction, completion, and applications: A literature study. Journal of Biomedical Informatics, 143:104403. DOI: 10.1016/j.jbi.2023.104403.

Rehman, S. U., Khan, A. U., and Fong, S. (2012). Graph mining: A survey of graph mining techniques. In Seventh International Conference on Digital Information Management (ICDIM 2012), pages 88–92. DOI: 10.1109/ICDIM.2012.6360146.

Yang, W., Some, L., Bain, M., and Kang, B. (2025). A comprehensive survey on integrating large language models with knowledge-based methods. Knowledge-Based Systems, 318:113503. DOI: 10.1016/j.knosys.2025.113503.

Zhu, Y., Wang, X., Chen, J., Qiao, S., Ou, Y., Yao, Y., Deng, S., Chen, H., and Zhang, N. (2024). Llms for knowledge graph construction and reasoning: recent capabilities and future opportunities. World Wide Web, 27(5):58. DOI: 10.1007/s11280-024-01297-w.