Towards a Knowledge Graph of Health Evolution

Morales Tirado, Alba Catalina; Daga, Enrico and Motta, Enrico (2022). Towards a Knowledge Graph of Health Evolution. In: Knowledge Engineering and Knowledge Management, Lecture Notes in Computer Science, Springer, Cham, pp. 105–120.

DOI: https://doi.org/10.1007/978-3-031-17105-5_8

Abstract

Electronic Health Records (EHR) contain detailed data of a person's health conditions and could provide emergency first responders with useful information. In previous works, we envisaged an intelligent system able to inspect health records and identify people in need of special assistance by reasoning on the evolution of conditions over time. Unfortunately, there is a lack of resources regarding health condition evolution and recovery time. However, information available on the web could help in supporting domain experts for building a database of Health Condition Evolution Statements (HES). This paper addresses this knowledge gap and proposes a four-step methodology based on knowledge acquisition (KA) techniques that support the extraction of HES from public sources. The approach uses text classification algorithms and exploits SNOMED CT taxonomy to build a database of HES. More importantly, the proposed KA pipeline includes a human-in-the-loop model that captures knowledge from experts and ensures the construction of high-quality Knowledge Graphs (KG) to support the task at hand. We evaluate the approach with domain experts' help and discuss the user study results. Finally, we contribute the first curated Knowledge Graph of HES.

Viewing alternatives

Metrics

Item Actions

Export

You can export this page using these formats Digital Object Identifier - DOI