Concept relevance metrics in the Medical Subject Headings (MeSH)

In the MeSH concept relevance dataset, we define metrics to quantify relevance within this knowledge organization system and have applied them to the MeSH hierarchy. The research uses the 2022 PubMed collection, which contains over 33 million articles, and builds citation networks from January 2014 to December 2021. We quantify the relevance of MeSH concepts based on disruptiveness, influence, informativeness, usefulness, and concept relevance, using metrics such as disruption, graph centrality, entropy, category utility, and reciprocal rank fusion.