The Human Phenotype Ontology, or HPO, supplies standardized terms for clinical abnormalities. Its value is not merely consistent spelling. A shared hierarchy lets software recognize that a specific observation belongs within a broader clinical concept, making phenotypes more useful for matching cases and prioritizing genomic candidates.

Illustrated overview of describe the phenotype precisely, highlighting specific terms, time and context, meaningful absence.
Describe the phenotype precisely: specific terms, time and context, meaningful absence. Read the article for evidence and limitations. Original conceptual illustration.

Choose the observation the evidence supports

The HPO project’s 2024 update describes an expanding vocabulary, disease annotations and work across languages and clinical systems. When using it, choose the most specific term supported by the record. Specificity should come from examination or investigation, not from assuming a suspected diagnosis is already correct. For example, a broad description of seizures should not become a particular seizure type without adequate documentation. Read the HPO resource paper.

Retain the original clinical wording alongside the standardized term when a distinction matters. This allows later reviewers to understand why the term was selected and to correct an overly narrow or inaccurate mapping. A code alone does not capture the confidence of the observation.

Represent time and negative findings

A feature can be present, absent after assessment, or simply unknown. Those states should not be interchangeable. Likewise, an age-dependent feature absent in infancy may have a different implication from the same finding absent after appropriate assessment in adulthood. Onset, progression and the timing of evaluation can therefore change the diagnostic meaning of an otherwise identical term.

Editorial illustration: A detailed observation notebook with distinct eye, ear and limb symbols, a magnifying glass and a timeline made of dated coloured dots.
Conceptual illustration accompanying this article. People and situations depicted are fictional; this is not a diagnostic image.

GA4GH Phenopackets offer a structured way to exchange phenotype information together with contextual data about an individual. They complement HPO rather than replace it: the ontology describes concepts, while the data format organizes observations and related information for exchange. Explore the Phenopackets standard.

A quality check before sharing

  • Confirm that each term describes the person rather than a family member or a differential diagnosis.
  • Distinguish directly observed findings from historical reports.
  • Check negation, age at onset and outdated or duplicate terminology.
  • Keep provenance so a reviewer can return to the underlying assessment.

These checks are especially valuable when software extracts phenotypes from notes. A fluent extraction can still confuse a ruled-out diagnosis with an established finding. Structured phenotyping works best as a maintained clinical record: concise enough to use, detailed enough to interpret, and revised when the person’s presentation changes.