A multi-factor approach to identifying missed synonymy in the UMLS

Motivation: In the Unified Medical Language System (UMLS), the eponymic term “Brodie abscess” denotes a chronic metaphyseal abscess of a bone, named after Sir Benjamin Collins Brodie, a 19thCentury British physician. Now, is “Brian abscess” also a term of the UMLS an abscess named after some Dr. Brian or a misspelled form of the term “Brain Abscess”? Similarly, the two terms “antidiarrheals” (hyphenated) and “antidiarrheals” (one-word) seem to denote the same pharmacological class, yet are separate concepts in the UMLS. Misspellings and, more generally, uncontrolled lexical variation in biomedical terminologies lead to missed synonymy and have detrimental consequences on data retrieval, reasoning or literature mining processes.