Translating medical vocabularies via the Unified Medical Language System William T. Roddy1, Daniel Olson,1, Diane Corey1, Ian Braun1, Terrence R. McHugh1, Emily Hartley1, Smith Heavner1, Ramona L. Walls1 1 Critical Path Institute, Tucson, AZ Keywords 1 data management, OMOP, SDTM, CDISC, UMLS, terminology, vocabulary 1. Background Outcomes Partnership (OMOP) model prior to data integration. The relevant concepts within the integrated data will be further mapped to OBO The Rare Disease Cures Accelerator – Data ontologies (https://obofoundry.org/) prior to and Analytics Platform (RDCA-DAP) of the knowledge graph ingestion (which is outside the Critical Path Institute (C-Path) is an FDA-funded scope of this submission). The OMOP effort to facilitate drug development for rare standardized vocabulary includes many diseases (https://portal.rdca.c-path.org/). RDCA- biomedical terminologies that enable DAP helps researchers leverage existing standardization of source data; however, these knowledge and analyze data to inform and terminologies do not currently include the Study optimize clinical trial design with new sources of Data Tabulation Model (SDTM) controlled evidence. The platform supports the use of data to terminology which is frequently used for improve the quantitative characterization of rare submissions to regulatory authorities. Generating disease progression, define novel biomarkers and mappings between the SDTM terminology and endpoints, and provides analytical tools to inform the OMOP standardized vocabulary will further the design of innovative trial protocols. expand the capabilities of data sharing and reuse One of the key deliverables is the creation of a between real-world data sources and clinical trial knowledge graph from the natural history, registry data sources. We demonstrate an implementation and clinical trial data received. The data, however, of translating the terminology used in these two must be cleaned and standardized prior to Common Data Models. knowledge graph ingestion which has presented us with opportunities to implement novel (to our organization) automation procedures of certain data management activities (e.g. vocabulary 1.1. Methods mappings). Recent advances in clinical research data We have developed mappings between standards have resulted in the development of SDTM terminology and the OMOP standardized several Common Data Models (CDMs) which are vocabulary by using the Unified Medical leveraged to support the sharing and reuse of Language System (UMLS). The SDTM data1. Critical Path Institute has chosen to map terminology is published by the National Cancer legacy data to the Observational Medical Institute Enterprise Vocabulary Services (NCI ICBO 2022, September 25-28, 2022, Ann Arbor, MI, USA EMAIL: wroddy@c-path.org (A. 1); dolson@c-path.org (A. 2); dcorey@c-path.org (A. 3) ibraun@c-path.org (A. 4) terrence.r.mchugh@gmail.com (A.5) ehartley@c-path.org (A.6) sheavner@c-path.org (A. 7) rwalls@c-path.org (A. 8) ORCID: 0000-0002-8453-520X (A. 1); 0000-0002-8134-1207 (A. 2); 0000-0003-3840-2315 (A. 3) 0000-0002-2389-9288 (A. 4) 0000-0002-3805-0359 (A.5) X (A.6) 0000-0003-0912-0407 (A. 7) 0000-0001-8815-0078 (A. 8) ©️ 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR Workshop Proceedings (CEUR-WS.org) CEUR ceur-ws.org Workshop ISSN 1613-0073 Proceedings EVS), where concepts are identified by concept mappings require further refinement based on codes (C-Codes) which are included within the subject-matter expert review and additional UMLS. We identified all UMLS Concept Unique transformation logic, to ensure that context Identifiers (CUIs) by searching for atoms (the appropriateness of mappings. For example, we smallest unit of naming in a source) with a are exploring further refinement by including the source abbreviation of NCI and a source code source SDTM domain in the mapping logic. containing the C-Code in the SDTM Additionally, it may be possible to bolster the terminology. The UMLS CUIs associated with mappings with additional resources such as the SDTM terminology were used to retrieve CDISC’s LOINC to LB Mapping Files. source codes originating from terminologies that are included within the OMOP vocabulary. We primarily focused on the SNOMED-CT, LOINC, RxNorm, and UCUM vocabularies because they are standard within the OMOP vocabulary, but we also expanded the search to MedDRA and MeSH due to their appreciable representation in SDTM and OMOP. The source codes from these vocabularies were used to identify the equivalent OMOP vocabulary standard concept. Figure 1: data flow with the number of unique C- 1.1.1. Status of Mapping Results Codes present in each The 2021-12-17 release of the SDTM terminology included 22,132 unique C-Codes. 2. Conclusion Of these, 84.2% were available within the UMLS 2021AB release and 57.4% were only indexed in We show that it is feasible to aid the an NCI EVS terminology or the Metathesaurus transformation process between CDMs by vocabulary. Within the UMLS searched utilizing the UMLS to generate mappings vocabularies 25.5% of the C-Codes were present between the SDTM terminology and the OMOP and 1.3% were present in other vocabularies. We vocabularies. Future work will expand and used all possible vocabulary codes to query the ensure accuracy of the mappings, outline OMOP vocabularies (release v5.0 28-JAN-22) improvements of data standards for and found that 19.2% of the C-Codes mapped to interoperability, and publish source code. a standard OMOP concept. There were no corresponding OMOP concepts for 4.6% of C- 3. References Codes; however, nearly 90% of these are UCUM concepts and this is expected based on the OMOP documentation2. [1] Garza M, Del Fiol G, Tenenbaum J, Walden To evaluate the applicability of this A, Zozus MN. Evaluating common data approach, we used SDTM data from the C-Path models for use with a longitudinal Online Data Repository3 to identify submission community registry. J Biomed Inform. 2016 values from controlled terminology codelists. We Dec;64:333-341. found that a majority of observations mapped to [2] Available from: at least one standard concept. The https://www.ohdsi.org/web/wiki/doku.php?i appropriateness of initial mappings was assessed d=documentation%3Avocabulary%3Aucum by comparing the SDTM codelist domain to the [3] Critical Path Institute Online Data target concept domain. In many cases the source- Repository (CODR). Available from: to-target domain were appropriate; for example, https://codr.c-path.org/ codes in the laboratory data (LB) domain had target concepts in the Measurement domain. This preliminary mapping shows potential in the ability to extract translations between SDTM and OMOP vocabularies through the UMLS. These