Modeling Matters: Applying UML to Pharma and IDMP Ontologies

Introduction: Why Modeling Is Non-Negotiable in Pharma

In the pharmaceutical industry, the stakes are high: complex products, global regulation, data integrity, and patient safety. Modeling is not a luxury — it’s a necessity. Unified Modeling Language (UML) is one of the foundational tools used to define, analyze, and standardize information flows across regulatory, clinical, and manufacturing domains.

One of the clearest examples of UML’s application in pharma is its use in supporting IDMP (Identification of Medicinal Products) — a set of ISO standards that require global harmonization of how medicines are described. IDMP uses formal ontologies to define medicinal products, substances, and packaging — and at the heart of those ontologies is structured UML modeling.

What Is IDMP?

IDMP stands for Identification of Medicinal Products , a suite of five ISO standards:

ISO 11615 – Medicinal product information
ISO 11616 – Pharmaceutical product information
ISO 11238 – Substances
ISO 11239 – Pharmaceutical dose forms, units of presentation, routes of administration, and packaging
ISO 11240 – Units of measurement

The goal is global interoperability — regulators (like EMA, FDA, PMDA) need to share structured, machine-readable data about products across jurisdictions. Ontologies are needed to describe and validate this complex domain. But those ontologies don’t just appear — they’re modeled first.

Why UML Before Ontology?

Ontologies express knowledge using logical formalisms (OWL, RDF, etc.), but they must be built on a solid structural foundation. That’s where UML comes in:

📘 UML defines concepts, attributes, relationships — the skeleton of any ontology
🔁 UML enables versioning and validation of models before publishing as formal ontology
🔍 UML class diagrams provide a readable, visual map for domain experts, analysts, and regulators
⚙️ Tools like Sparx EA export UML models to OWL/RDF or XML for transformation into semantic models

Example: UML Model for IDMP Substance

Let’s look at how UML is used to define the “Substance” domain in IDMP:

Substance is a class with attributes like Substance ID , Name , Structure , and Source
Substance may be Manufactured or Natural
Substance links to Reference Information and Controlled Vocabulary classes

This UML model forms the basis of:

Regulatory forms and submissions (SPOR / EMA)
Data exchange standards (HL7 FHIR, ISO XML)
Ontology definitions (OWL classes, restrictions)

Diagram Snapshot (Conceptual)

+----------------+       +------------------+
|   Substance    |<>-----|  SubstanceType   |
+----------------+       +------------------+
| - ID           |       | - TypeCode       |
| - Name         |       +------------------+
| - Structure    |
+----------------+

Without UML Modeling, Progress Stalls

Why is modeling mandatory?

❌ Without UML, no standard vocabulary or structure
❌ No validation or simulation of relationships
❌ Impossible to auto-generate schemas or APIs
❌ Regulatory submissions become manual and inconsistent
❌ Ontology teams struggle to align with business experts

UML bridges the world of structured modeling and semantic representation. It's the prerequisite for traceability, automation, and validation .

Client Example: UML-Driven IDMP Implementation

We supported a multinational pharma firm in designing their IDMP data platform. The core design artifact? UML models for:

Substances, Products, and Authorizations
Dosage forms and units of presentation
Controlled vocabularies and code lists

We used Sparx EA to model these domains, apply business rules (using OCL and stereotypes), and export structured definitions for ontology builders. The result:

✅ 100% traceable data from design to implementation
✅ Schema reuse across EMA and FDA requirements
✅ Faster onboarding for new regulatory standards

Conclusion: UML Is the Foundation of Semantic Pharma

In pharma, data is a product. That product must be modeled, validated, and exchanged globally. UML isn’t optional — it’s the foundation. From IDMP ontologies to regulatory data lakes, modeling brings order to biological and chemical complexity.

Before you can build the ontology, define the structure. That’s UML’s job — and why it’s more critical than ever.