2026-04-15: DEUSG Vienna

Part 1:

Drug Model Implementation Guide Updates

The meeting focused on updates to the drug model implementation guide and dose form modeling. Alejandro presented a new draft implementation guide that covers use cases, content structure, and technical application, which is available for feedback in the forums. The group discussed two options for consolidating intended site and route of administration value sets, with one proposal suggesting a shared common site approach and another recommending expansion of existing hierarchies. The meeting also addressed dose form after transformation modeling, exploring options for explicit representation of transformed dose forms and their implications for classification and technical implementation.

Pharmaceutical Dose Form Modeling Approaches

The discussion focused on modeling approaches for product transformation information in pharmaceutical dose forms. The team explored several options including a flat model approach and additional relationships to better capture information about transformations from manufactured forms to administrable forms. They discussed how to handle cases where a single manufactured dose form can be transformed into multiple administrable forms, and considered the alignment between dose form classification and medicinal product classification. The conversation highlighted potential challenges in querying specific administrable forms while maintaining consistency in modeling across different product types.

Administrable Dose Form Modeling Challenges

The team discussed challenges with modeling administrable dose forms, particularly the distinction between injection and infusion products. They identified that the current model may not accurately represent the hierarchical relationship between different dose forms and could benefit from additional attributes to better capture how products are classified and searched. The discussion also touched on differences in strength representation between presentation strength and concentration strength, noting that most SNOMED content uses concentration strength rather than presentation strength. The team concluded that a separate administrable dose form property at the product level would be helpful for better alignment with IDMP requirements and prescription support.

Pharmaceutical Product Dose Form Challenges

The team discussed challenges around representing pharmaceutical products in different dose forms, particularly powders that can be transformed into solutions. They explored how to handle situations where products can exist in multiple forms (powder vs. solution) and the potential for confusion in the system. The discussion concluded that new attributes would be needed to support both clinical and pharmaceutical perspectives, with flexibility to handle different administration methods, though the specific modeling approach (nested or otherwise) remains to be determined.

Dose Form Model Representation Discussion

The discussion focused on representing dose forms in a model, specifically whether to include administrable dose form information as part of the minimum sufficient set for subsumption. The participants debated whether this additional information should be required for classification or kept as a separate axiom. It was noted that the EDQM model already includes this information, and aligning with international standards like those in Europe, USA, and Health Canada would benefit regulators. The group concluded that a compromise approach using dose form after transformation as a flat model would allow for easier maintenance while still providing the necessary information for querying medicinal product models.

Dosage Form Modeling Attributes Discussion

The team discussed modeling options for administrable dosage forms, focusing on whether to use a single attribute or two parallel attributes to capture different forms after transformation. They agreed that having two attributes with different domains would be more aligned with their requirements and easier to implement, though not mandatory. The group also considered the hierarchical relationship between transformation types and decided to seek group review on potentially changing the current hierarchy arrangement.

Administration Form Transformation Challenges

The team discussed challenges with handling concepts that transform into different administration forms, specifically focusing on a powder concept that transforms into both intravesical solution and solution for injection. They agreed that role grouping would be necessary for this edge case where administration methods and intended sites differ, though this approach was noted as a temporary solution rather than a permanent modeling change. The group decided to focus on testing Option 4 and verify if it would cause issues with the Canadian Extension before making a final decision. 

Part 2:

Substances to Organisms Linking Challenges

Yong presented an update on the substances linked to organisms project, focusing on internal testing and potential challenges. The key challenges identified include the lack of a clear definition for substances, difficulties in classifying substances due to the manual maintenance of the hierarchy, and issues with linking substances to organisms, particularly in the context of taxonomy and classification changes. Young demonstrated these challenges using an example of cannabis-related substances and organisms, showing that many substances are not exclusive to a specific organism and highlighting the limitations of using AI for automatic linking. The discussion also touched on philosophical debates regarding the definition of substances, such as whether a statue made of clay should be considered one or two entities.

SNOMED CT Substance Definition Framework

The discussion focused on defining substance in SNOMED CT and establishing a formal ontological framework. The speaker explained that substance should be defined as a material entity that is a collective or compound of material entities rather than a simple collection, with specific emphasis on defining constituents present in sufficient proportion. The team discussed various attribute names for linking substances to organisms, including "source organism," "derives from," and "source material," with particular attention to ensuring the defining relationships are appropriate for their use cases. The conversation concluded with a discussion about substance descriptions and potential duplicates with other hierarchies like organism or product.

Cannabis Substance Terminology Discussion

The discussion focused on handling cannabis substance terminology in a semantic hierarchy. The speaker outlined four options for describing cannabis substances, including keeping the current "cannabis substance" terminology, renaming it to "cannabis material," using "substance" in descriptions, or specifying "substance derived from cannabis." The speaker noted that current hierarchical arrangements may need changes depending on whether the concept is interpreted as generic or specific to drugs/medicinal products. The discussion also addressed the ongoing issue of inactivating proxy concepts in regular releases, which has caused problems for NLCs, and the speaker recommended stopping these activations until an agreement is reached, as proxies are necessary for upper-level classes in the product hierarchy.

Cannabis Substance Classification Challenges

The discussion focused on challenges in implementing a generic proxy concept for substances derived from cannabis and other organisms. The participants debated the logical structure and hierarchy of substance classifications, particularly around whether substances from manufactured products should be subsumed under plant materials. They identified issues with current terminology like "plant product" being ambiguous between natural plant products and manufactured products. The conversation explored different classification approaches, including using source materials and organism hierarchies, while highlighting concerns about maintaining logical consistency and avoiding overly broad generalizations that could compromise the system's utility.

Substance Hierarchy Classification Challenges

The team discussed challenges with the substance hierarchy, particularly regarding how to categorize plant parts and materials. They explored issues with the current substance classification system, including problems with collective vs. individual representations and the complexity of categorizing natural materials. The discussion highlighted concerns about how changes might affect existing systems, particularly regarding cannabis modeling where approximately 10-20 disorder concepts currently use cannabis substance. The team debated whether to maintain the existing cannabis substance concept while fixing parent-child relationships, or to create new explicit concepts, with considerations for minimizing disruption to international models.

Cannabis Oil Modeling Challenges

The team discussed challenges with modeling cannabis oil substances and products in their system. Matt and Yong debated whether cannabis oil should be treated as a substance or product, with Matt arguing it should remain a product rather than being modeled as a medicinal substance. Guillermo raised concerns about the fundamental distinction between substances (which can be measured and divided) and products (which are countable entities), suggesting the current substance hierarchy needs significant rethinking. The discussion was time-boxed due to other priorities, with the team acknowledging that while the current approach may have limitations, they need to balance precision with practical implementation given the volume of natural products that need to be modeled.

Drug Ingredient Representation Standards

The group discussed two main topics regarding drug ingredient representation. First, they addressed how to handle clinical drugs with inactive ingredients, deciding to use the "count of active ingredient zero" approach rather than creating a new "has no active ingredient" attribute. Second, they discussed rounding algorithms for concentration normalization, agreeing to maintain the current Canadian algorithm of 3 decimal places rather than switching to the international release's method. The conversation ended with a discussion about meeting cadence, deciding to maintain fortnightly meetings at 11 UTC, and identified the need for future discussions about different modeling patterns across countries, particularly regarding presentation strength for single dose products.