UMLS Marks 40 Years of Advancing Biomedical Data Integration

The Unified Medical Language System (UMLS), a foundational National Library of Medicine (NLM) resource for integrating and connecting biomedical and health information, marks its 40th anniversary in 2026. Initiated in 1986 and first released in 1990, the UMLS was developed to help computer systems understand and integrate information expressed using different biomedical vocabularies and coding systems. At its core, the UMLS brings together terminology from many sources and connects terms that have the same or related meanings, enabling information recorded using different vocabularies to be linked and interoperable.

Over four decades, the UMLS has grown substantially in both scale and use. The UMLS 2026AA Release contains 3,530,466 active concepts and 18,064,970 concept names drawn from 195 source vocabularies. The UMLS is a concept-based integration model that assigns Concept Unique Identifiers (CUIs) to biomedical concepts and brings synonymous terms from different source vocabularies together under a common concept. At the same time, the UMLS maintains links to the original source terms and codes, preserving their provenance while enabling connections and mappings across terminologies. Source vocabularies are incorporated through ongoing acquisition and integration processes coordinated with external terminology providers, with updates consolidated and distributed through semiannual UMLS releases. Because these sources include both publicly available and proprietary terminologies with differing licensing requirements, the UMLS also provides a controlled-access framework for managing licenses and enabling users to access and use terminology content. Through the UMLS Terminology Services (UTS), users can search and browse UMLS content, access terminology through application programming interfaces (APIs), and download UMLS data for use in their own systems and applications.

Use of UMLS services has increased markedly as biomedical information has become increasingly digital and interconnected. Annual UMLS API use grew from approximately 25 million calls in FY2016 to more than 428 million calls in FY2025—a nearly 17-fold increase. In the first three quarters of FY2026 alone, the UMLS API served nearly 399 million calls. Users also downloaded UMLS data more than 8,300 times in FY2025, while the UMLS and UTS websites together recorded more than one million page views. These measures reflect the continuing importance of programmatic and downloadable access to UMLS terminology resources.

The UMLS has continually evolved alongside changes in biomedical informatics and information technology. Following its first release in 1990, UMLS production transitioned in the late 1990s from primarily batch-oriented processing toward a more sophisticated editorial workflow. Distribution similarly evolved from physical media such as CD-ROMs to downloadable electronic files. Beginning in the late 1990s and early 2000s, web-based tools, including the UMLS Knowledge Source Server, gave users new ways to search, navigate, and explore UMLS concepts, relationships, and source vocabularies. The introduction of the UMLS REST API in 2015 further expanded machine-readable access and enabled developers to more easily test integration of UMLS content directly into applications and computational workflows.

The UMLS has also supported advances in biomedical natural language processing. Beginning in the 1990s, NLM developed the SPECIALIST Lexicon and associated lexical tools to support computational processing of biomedical language. These capabilities were complemented by tools such as MetaMap, which maps biomedical text to UMLS concepts, and SemRep, which identifies semantic relationships in biomedical text. Together, these resources helped establish the UMLS as an important foundation for extracting, connecting, and interpreting information from biomedical literature and other text. The UMLS continues to serve as a bridge from modern AI-based NLP tools to standard concepts and identifiers.

Throughout its history, the UMLS has helped researchers, developers, terminology specialists, and health information systems bridge differences among vocabularies. By linking concepts across terminology and coding systems, the UMLS supports applications ranging from information retrieval and natural language processing to clinical research, data integration, terminology mapping, and interoperability. It also provides an important foundation for other NLM terminology resources, including RxNorm.

As the UMLS enters its fifth decade, NLM is working to modernize how UMLS content is curated, produced, and distributed while preparing the resource for emerging artificial intelligence use cases. This evolution will include exploring more modern approaches to terminology curation, release formats, and tooling to make UMLS knowledge easier to use in contemporary computational environments. Building on 40 years of connecting biomedical knowledge, the UMLS will continue to evolve to help researchers, developers, and health information systems integrate diverse data and make biomedical information more computable, interoperable, and useful for research and health.