Pulse Brain · Growing Health Evidence Index
Tier 3 — Observational / field trialPeer-reviewed

PheKB: a catalog and workflow for creating electronic phenotype algorithms for transportability

Jacqueline Kirby, Peter Speltz, Luke V. Rasmussen, Melissa Basford, Omri Gottesman, Peggy Peissig, Jennifer A. Pacheco, Gerard Tromp, Jyotishman Pathak, David Carrell, Stephen B. Ellis, Todd Lingren, Will K Thompson, Guergana Savova, Jonathan L. Haines, Dan M. Roden, Paul A. Harris, Joshua C. Denny

Journal of the American Medical Informatics Association · 2016

Read source ↗ All evidence

Summary

This paper reports on the Phenotype KnowledgeBase (PheKB), an online platform enabling researchers to build, share, and validate electronic phenotype algorithms for mining electronic health records. As of June 2015, the repository contained 30 finalized and 62 in-development phenotype algorithms with demonstrated high transportability across institutions: median case PPV was 96.0% at authoring sites and 97.5% at implementation sites, demonstrating that algorithms developed at one institution generalise effectively to others. The findings support PheKB's utility as a central repository for developing research-grade phenotypes from health care data.

UK applicability

The methodology and findings regarding phenotype algorithm development and transportability are broadly applicable to UK NHS and health research settings, particularly for researchers working with NHS electronic health record data. However, direct adoption would require adaptation to UK coding systems (e.g. SNOMED CT) and NHS-specific data structures, and UK researchers would benefit from equivalent UK-based repositories.

Key measures

Positive predictive value (PPV) for cases and controls at authoring vs. implementation sites; algorithm component frequency (ICD codes, medications, natural language processing); phenotype reuse metrics

Outcomes reported

The study documented the development, sharing, and validation of 30 finalized electronic phenotype algorithms and 62 in-development algorithms across multiple health care systems. Performance metrics including positive predictive values (PPVs) for case and control identification were compared between authoring institutions and secondary implementation sites.

Theme
Measurement & metrics
Subject
Measurement methods & nutrient profiling
Study type
Research
Study design
Descriptive report with implementation validation
Source type
Peer-reviewed study
Status
Published
Geography
United States
System type
Human clinical
DOI
10.1093/jamia/ocv202
Catalogue ID
BFmovi1onn-o0njvi

Topic tags

Pulse AI · ask about this record

Dig deeper with Pulse AI.

Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.