Ontology inventory
A semantic foundation for regulated scientific execution.
The ontology catalog includes open, enterprise, and licensed vocabularies that can be combined with internal taxonomies while preserving source, version, license, and permitted use.
Ontology domains
Coverage across the scientific and regulated operating model.
Clinical terminology and outcomes
Support diagnoses, procedures, observations, medications, outcomes, and care context.
- SNOMED CT
- ICD-10
- LOINC
- RxNorm
- HCC and clinical classifications
Adverse events, product risk, and submissions
Normalize safety concepts, regulatory terms, controlled terminology, and reporting structures.
- MedDRA through BYOL where required
- CTCAE
- CDISC Controlled Terminology
- NCIt
- Regulatory product concepts
Products, substances, classes, and mechanisms
Connect ingredients, products, classifications, molecular entities, and pharmacologic relationships.
- ATC
- ChEBI
- RxNorm
- DrugBank through BYOL
- CVX and product vocabularies
Genes, variants, proteins, pathways, and disease biology
Resolve molecular entities and relationships for discovery, translational, and biomarker workflows.
- HGNC
- Gene Ontology
- UMLS mappings
- COSMIC through BYOL
- MeSH and NCIt disease concepts
Measurements, tests, procedures, and findings
Standardize laboratory observations, imaging concepts, units, methods, and result context.
- LOINC
- RadLex
- UCUM-aligned units
- Assay and method taxonomies
- Imaging and biomarker concepts
Literature, methods, study design, and evidence objects
Connect research questions, designs, populations, endpoints, publications, methods, and evidence claims.
- MeSH
- NCIt
- CDISC terminology
- Internal evidence taxonomies
- Study and endpoint models
Semantic capabilities
Ontologies become executable services, not static reference files.
Concept and synonym resolution
Map acronyms, aliases, brand names, historical terms, local codes, and spelling variants to governed concepts.
Cross-ontology relationship mapping
Connect equivalent, broader, narrower, related, and context-dependent concepts across vocabularies.
Taxonomy and graph traversal
Use parent-child and relationship structures for search, cohort logic, aggregation, inference, and impact analysis.
Version-aware meaning
Preserve which ontology version applied, when a term changed, what superseded it, and which decisions used it.
Confidence, ambiguity, and review
Record automatic, suggested, reviewed, accepted, rejected, and exception mappings with named authority.
Internal vocabularies and local semantics
Federate corporate taxonomies, product hierarchies, proprietary biomarkers, assay terms, and operating classifications.
Ontology lifecycle
Every ontology and mapping follows a governed lifecycle.
STEP 01
Register the source
Ontology, steward, license, release, scope, provenance, and deployment eligibility.
STEP 02
Load and validate
Concepts, hierarchy, synonyms, relationships, identifiers, integrity, and completeness.
STEP 03
Map enterprise meaning
Canonical entities, crosswalks, local terms, confidence, ambiguity, and review status.
STEP 04
Approve operational use
Module, role, geography, purpose, model, workflow, external-use status, and authority.
STEP 05
Monitor changes
New releases, retired concepts, hierarchy changes, licensing changes, and downstream impact.
STEP 06
Preserve lineage
Version, mapping, decision, review, supersession, and Evidence Data Trace.
Version and mapping control
Meaning is governed as carefully as data.
A term can change meaning across markets, time periods, source systems, products, and regulatory contexts. StromaOS preserves ontology version, local mapping, review status, confidence, intended use, and downstream dependencies.
This allows agents and workflows to distinguish exact equivalence from approximate similarity and to escalate unresolved semantic ambiguity before it affects evidence or decisions.
Model an ontology federationLicensing and deployment
Use open and licensed vocabularies without compromising control.
Included where licensing permits
Open vocabularies can be registered, versioned, mapped, and deployed as governed platform services.
Licensed assets remain customer-controlled
Restricted ontologies such as MedDRA, DrugBank, or COSMIC can be invoked under the customer's existing license and deployment terms.
Cloud, VPC, and air-gapped operation
Ontology services can run inside customer-controlled environments with no external data movement.
Purpose- and role-aware access
The Sentry Agent checks license, market, role, intended use, and external-use restrictions before execution.
Governance inside the Fleet
The Sentry Agent governs ontology use at runtime.
Governance controls are encapsulated inside the Fleet. The Sentry Agent verifies that the correct ontology version, mapping, license, source, and permitted use apply before Lead and Worker Agents can rely on a concept.
Use the correct release for the decision context.
Ontology versions remain linked to time, market, study, submission, analysis, and evidence record.
Prevent unauthorized ontology access or redistribution.
BYOL assets remain restricted to licensed environments, users, workflows, and purposes.
Escalate uncertain or consequential mappings.
Low-confidence, one-to-many, context-dependent, and unmapped concepts require review or human adjudication.
Preserve how meaning affected evidence and decisions.
Every resolved term, hierarchy traversal, mapping, query, and downstream use remains traceable.
Platform outcomes
Scientific meaning becomes consistent, portable, and defensible.
Resolve different codes into shared meaning.
Agents and workflows can reason across vendors, markets, systems, and data models without repeated manual interpretation.
Build cohorts and evidence using explicit concept logic.
Preserve included concepts, exclusions, hierarchy depth, mapping confidence, version, and review.
Manage terminology evolution without breaking workflows.
Track retired concepts, new releases, remappings, and downstream impact before operational use.
Give every AI agent the right scientific vocabulary.
Agents receive only the ontology services, versions, mappings, and licensed assets permitted for their task.
Start with one domain where terminology mismatch slows analysis, evidence review, or cross-system execution.
Register the vocabularies, map enterprise terms, define version and license controls, configure Sentry governance, and expose the ontology as a shared execution service.