OHDSI Standardized Vocabularies-a large-scale centralized reference ontology for international data harmonization

Area of Science:

• Health Informatics
• Observational Research
• Ontology Engineering

Background:

• The Observational Health Data Sciences and Informatics (OHDSI) network is the world's largest, with over 331 data sources and 2.1 billion patient records across 34 countries.
• Large-scale observational research requires standardized data through a common data model (CDM), necessitating a robust ontology system for data harmonization.

Purpose of the Study:

• To create and implement a comprehensive, efficient, and reliable ontology system for the OHDSI network.
• To enable data harmonization and support large-scale observational research by standardizing data representation.

Main Methods:

• Developed the OHDSI Standardized Vocabularies, a mandatory reference ontology for all OHDSI network data sites.
• Integrated imported and de novo-generated ontologies, defining concepts, relationships, and conversion processes to the Observational Medical Outcomes Partnership (OMOP) CDM.
• Ensured harmonization through assigned clinical domains, comprehensive entity coverage, support for international coding schemes, and standardization of semantically equivalent concepts.

Main Results:

• The OHDSI Standardized Vocabularies contain over 10 million concepts from 136 vocabularies.
• The system has been downloaded over 50,000 times by more than 8,600 users and is utilized by numerous research groups and data networks.
• This open-source resource effectively addresses challenges in large-scale observational research, facilitating efficient phenotyping, covariate construction, and various prediction and estimation tasks.

Conclusions:

• OHDSI provides an unparalleled, open vocabulary system crucial for global observational research.
• Researchers are encouraged to utilize this dynamic resource and contribute their use cases to its ongoing development.