Observed Phenotypes and Candidate Ontology Mappings

How phenotype pipelines can preserve the source observation separately from ontology normalization, contextual qualifiers, and competing mappings. The distinction matters because uncertain or negated concepts can alter patient similarity and downstream classification, while concept-mapping studies vary in evaluation and external validation. This topic separates two questions: what the record asserts, and which ontology concept best represents it. Mapping disagreement changes coding only if the original observation and its qualifiers remain available as a distinct, immutable field. Otherwise, normalization can change analytic meaning. Contributions should compare primary exports based on confirmed observations with separately identified sensitivity datasets containing candidate mappings.