Creating a next-generation phenotype library: the health data research UK Phenotype Library

Area of Science:

• Health Informatics
• Biomedical Research
• Data Science

Background:

• Reproducible research is crucial for advancing health data utilization.
• Existing platforms lack comprehensive tools for managing and sharing electronic health record (EHR) phenotyping algorithms.
• Standardization and accessibility of phenotyping algorithms are needed to scale health research.

Purpose of the Study:

• To develop a platform enabling health data users to find, access, curate, and reuse EHR phenotyping algorithms at scale.
• To create a centralized, accessible resource for managing and sharing complex health data analysis tools.

Main Methods:

• Engaged stakeholders to define platform requirements.
• Designed a web application incorporating a novel metadata standard for phenotyping algorithms.
• Implemented features for API access, computable data flows, version control, and direct researcher submission.

Main Results:

• Launched the Phenotype Library in October 2021, hosting 1049 phenotype definitions across 40 data sources and over 200,000 terms.
• Demonstrated utility through case studies supporting research hubs and informatics tool development.
• The platform provides structured metadata management, API access, and support for computable phenotypes.

Conclusions:

• The Phenotype Library is the first openly available, programmatically accessible resource for managing phenotyping algorithms globally.
• Removing barriers to sharing and computing phenotypes will unlock the potential of health data for public benefit.
• This initiative promotes reproducible research and accelerates discoveries in patient and public health.