UC Merced

We leverage the knowledge network representation of the Medical Subject Heading (MeSH) ontology to infer conceptual distances between roughly 30,000 distinct MeSH keywords — each being prescribed to particular knowledge domains — in order to quantify the origins of cross-domain biomedical convergence. Analysis of MeSH co-occurrence networks based upon 21.6 million research articles indexed by PubMed identifies three robust knowledge clusters: micro-level biological entities and structures; meso-level representations of systems, and diseases and diagnostics; and emergent macro-level biological and social phenomena. Analysis of cross-cluster dynamics shows how these domains integrated from the 1990s onward via technological and informatic capabilities — captured by MeSH belonging to the “Technology, Industry, and Agriculture” (J) and “Information Science” (L) branches — representing highly controllable, scalable and permutable research processes and invaluable imaging techniques for illuminating fundamental yet transformative structure–function–behavior questions. Our results indicate that 8.2% of biomedical research from 2000 to 2018 include MeSH terms from both the J and L MeSH branches, representing a 291% increase from 1980s levels. Article-level MeSH analysis further identifies the increasing prominence of cross-domain integration, and confirms a positive relationship between team size and topical diversity. Journal-level analysis reveals variable trends in topical diversity, suggesting that demand and appreciation for convergence science vary by scholarly community. Altogether, we develop a knowledge network framework that identifies the critical role of techno-informatic inputs as convergence bridges — or catalyzers of integration across distinct knowledge domains — as highlighted by the 1990s genomics revolution, and onward in contemporary brain, behavior and health science initiatives.