UMLS mapper finds biomedical concepts in Spanish clinical texts. Specifically, it finds concepts included in the UMLS Metathesaurus, which is a huge collection of biomedical terminologies of reference, such as SNOMED CT® and the Medical Subject Headings (MeSH). UMLS mapper is highly configurable: users can choose the types of concepts the tool must find, as well as the terminologies that it must use. Furthermore, the different components that make the UMLS mapper pipeline can also be configured.
UMLS mapper has proven to be robust across text genres (i.e., physician notes, clinical cases, scientific articles), and to perform in line with long existing and robust tools for English, such as MetaMap. An evaluation of UMLS mapper against the Mantra Gold Standard Corpus has yielded F1 measures for concept recognition, concept classification and concept linking of 0.70, 0.66 and 0.63, respectively.
UMLS mapper is developed at Vicomtech in collaboration with the IXA research group of the University of the Basque Country (UPVH/EHU). The work has been partially financed by the Department of Economic Development and Infrastructure of the Basque Government under the project BERBAOLA (KK-2017/00043), and by the Spanish Ministry of Economy and Competitiveness (MINECO/FEDER, UE) under the projects CROSSTEXT (TIN2015-72646-EXP) and TUNER (TIN2015-65308-C5-1-R).