Expand curated SNOMED thesaurus beyond 49 terms

[AlexMikhalev]

Summary

data/snomed_thesaurus.json has only 49 hand-curated terms covering oncology/PGx. Using UMLS account, systematically extract and curate terms for additional clinical domains.

Details

• Current 49 terms cover oncology and pharmacogenomics well
• High-value expansion domains: cardiology, neurology, endocrinology, infectious disease
• Each term needs proper SNOMED CT concept ID and preferred term
• Curated terms feed the knowledge graph with high-precision entities

Acceptance Criteria

• 500+ curated terms with SNOMED CT concept IDs
• Coverage across at least 5 clinical domains
• Validated against UMLS source for concept accuracy
• Entity extraction quality maintained or improved on 18 eval cases

Priority: P2