Common Core Ontologies
Home/Blog/Linking CCO and FHIR for Cross-Domain Health Data Integration
APPLIED

Linking CCO and FHIR for Cross-Domain Health Data Integration

November 20, 2026 · Marcus Webb
Linking CCO and FHIR for Cross-Domain Health Data Integration

FHIR standardizes how clinical systems exchange patient data, but it was never meant to be a general-purpose ontology. It has no formal answer for devices, supply chains, or research provenance that touch the same patients and processes. Common Core Ontologies fills that gap, giving teams one consistent semantic layer across clinical and non-clinical data alike.

Health systems rarely run on clinical data alone. A hospital tracks patient encounters through its EHR, but it also tracks infusion-pump maintenance schedules, cold-chain provenance for vaccine shipments, consent records for a research cohort, and staffing rosters for the unit that treated a given patient. All of that touches the same underlying entities (a patient, a device, an event), yet it is scattered across systems that were never designed to share a common vocabulary.

HL7 FHIR solved the clinical half of that problem well. It gives systems a shared resource model for exchanging Patient, Observation, Condition, and dozens of other clinical structures. Teams planning broader EHR integration services increasingly run into FHIR's edge almost immediately: the standard is clinically scoped by design, and it stops short of the equipment, supply-chain, and research-provenance data that sits right next to the clinical record it governs.

The rest of this article walks through what FHIR actually standardizes, where its scope ends, and how a mid-level ontology like Common Core Ontologies (CCO), built on Basic Formal Ontology (BFO), gives that non-clinical data a formal home without requiring a second clinical standard.

What FHIR Actually Standardizes (and Where It Stops)

HL7 FHIR (Fast Healthcare Interoperability Resources) defines a set of "Resources", including Patient, Observation, Condition, and MedicationRequest, as a shared data model for clinical exchange. Two systems that both speak FHIR can trade a patient's problem list or lab results without a bespoke integration for every pair of vendors. That is a real, hard-won interoperability win, and it is the reason FHIR has become the default clinical exchange layer across much of the industry.

What FHIR does not attempt is to be a general upper ontology. Its Resources are purpose-built for clinical exchange, with terminology bindings to code systems like SNOMED CT and LOINC rather than a domain-neutral class hierarchy. A FHIR Device resource can describe an infusion pump's identifier and type, but FHIR has no built-in way to reason about that pump as an Artifact with a maintenance Process, a manufacturer Agent, and a supply-chain history. That reasoning sits outside what a clinical exchange standard is scoped to do.

The Cross-Domain Gap: Where Clinical Data Meets Everything Else

This gap shows up anywhere a clinical system's data needs to connect to a non-clinical record that references the same entities. Three cases recur often enough to be worth naming directly.

Devices, Supply Chain, and Equipment Lifecycle Data

A FHIR Device resource identifies a piece of equipment used in care, but the maintenance history, calibration records, and supply-chain provenance behind that device typically live in a separate asset-management or procurement system. Without a shared semantic layer, a hospital cannot easily ask "which patients were treated with a device from this recalled manufacturing batch": the clinical and supply-chain records simply don't share a vocabulary that lets the question be posed as a single query.

Research Data and Provenance Tracking

A FHIR ResearchStudy resource can describe a clinical trial at a high level, but the detailed provenance a research team needs (which dataset version a result came from, which consent record authorized its use, which processing pipeline transformed it) is a lineage problem that FHIR's resource model was not built to express in depth. Research data provenance is its own well-studied problem, and it needs a formal structure that can sit alongside, not inside, the clinical resource.

Administrative and Operational Entities FHIR Wasn't Built to Model

Staffing assignments, facility scheduling, billing workflows, and the organizational structure of the care teams involved in a patient's treatment are operational realities that intersect clinical data constantly, but they are administrative concerns FHIR deliberately leaves to other systems. A hospital's EHR knows who treated a patient on a given visit; it rarely has a formal model connecting that fact to the broader organizational process that scheduled the encounter in the first place.

How CCO Complements FHIR

Common Core Ontologies is a mid-level ontology suite that extends Basic Formal Ontology (BFO), a domain-neutral upper ontology covering the most general categories of reality: objects, processes, qualities, and information entities. CCO specializes those top-level BFO categories into reusable mid-level classes, including Agent, Artifact, Event, and Information Content Entity, that are concrete enough to be useful across a wide range of domains without being tied to any one of them.

That positioning is exactly what the cross-domain gap needs. FHIR gives precise, clinically-scoped resources for exchanging patient data; CCO gives a domain-neutral vocabulary for reasoning about everything else those resources touch, from the device that delivered a treatment to the process that scheduled it. Mapped conceptually, a FHIR Patient resource corresponds to an instance of CCO's Agent class; a FHIR Observation corresponds to an Information Content Entity recording a measurement; a FHIR Device corresponds to an Artifact with its own lifecycle Process. None of this requires replacing FHIR. It requires treating FHIR's resource model as one well-defined source that feeds into a broader ontology-backed graph.

CCO does not compete with FHIR for clinical exchange. It sits one level up, giving the clinical resources FHIR already defines a place inside a larger, domain-neutral structure that also covers the equipment, supply-chain, and research data a hospital system has to track alongside them.

A Practical Pattern for Cross-Domain Interoperability

Teams that have gone through this exercise tend to converge on the same general pattern, regardless of which specific FHIR resources or CCO classes they end up using.

  1. Identify the FHIR resources already in production use. Typically Patient, Observation, Condition, and Device cover most of a clinical system's export surface.
  2. Map each resource type to a corresponding CCO class conceptually. Agent for people, Information Content Entity for records and measurements, Artifact for equipment, Process for the events that connect them.
  3. Extend the graph with the non-clinical entities FHIR doesn't cover. Supply-chain records, research provenance, and administrative processes get modeled as CCO classes directly, using the same relations (has agent, has input, precedes) that link them back to the mapped clinical entities.
  4. Resolve entities across systems through shared identifiers. A patient, device, or process referenced in both a FHIR resource and a non-clinical record needs a consistent identity so a query can follow the relation across the boundary.

The result is not a replacement for FHIR-based clinical exchange, but a layer above it: a graph where a device's supply-chain history, a patient's clinical record, and a research study's provenance chain can all be queried through the same relations, because they were all mapped into the same mid-level vocabulary rather than three disconnected schemas.

Frequently Asked Questions

Is FHIR itself an ontology?

No. FHIR defines a resource model, including Patient, Observation, Condition, and similar structures, for exchanging clinical data between systems, plus terminology bindings to code systems such as SNOMED CT and LOINC. It does not define a formal, domain-neutral class hierarchy with axioms a reasoner can check, which is what an upper or mid-level ontology like BFO or CCO provides.

Do I need CCO if I already use FHIR?

You need it only when your integration work extends past clinical exchange into domains FHIR was not built to model, such as equipment lifecycle and supply-chain provenance, research data lineage, or administrative and operational entities that reference the same patients and processes. For pure clinical-system-to-clinical-system exchange, FHIR alone is usually sufficient.

What's the difference between an upper ontology and a data exchange standard?

A data exchange standard like FHIR specifies the shape of messages and resources so two systems can parse each other's data correctly. An upper ontology like Basic Formal Ontology specifies the most general categories of reality, such as objects, processes, qualities, and information entities, so that classes from otherwise unrelated domains can be reasoned about consistently. One governs syntax and clinical semantics; the other governs cross-domain semantic structure.

Can FHIR resources be mapped to CCO classes directly?

Conceptually, yes. A FHIR Patient resource corresponds to an instance of an Agent-type class; a FHIR Observation corresponds to an instance of an Information Content Entity recording a measurement; a FHIR Device resource corresponds to an Artifact. The mapping is illustrative rather than a published one-to-one standard, so teams define the specific correspondence for their own data model.

Does this apply outside healthcare?

Yes. The same pattern, a domain-specific exchange standard handling one slice of the data paired with a mid-level ontology for cross-domain reasoning, shows up anywhere an industry standard is scoped to one function. CCO itself is domain-neutral; health data interoperability is one of several places teams apply it alongside a narrower standard.

What tools support CCO-FHIR mapping work today?

CCO is published as OWL and queried with SPARQL against a triplestore or graph database, the same standard tooling used for any ontology-backed dataset. FHIR resources are typically exported or transformed into RDF before this mapping happens; there is no single off-the-shelf tool that performs the CCO-FHIR correspondence automatically, so most teams build a transformation layer specific to the resources they actually use.

CCO and BFO are published as open ontologies registered with the OBO Foundry, which reviews them for logical consistency before adoption.