• Recording of meeting approved by participants.

Conflicts of interest and agenda review

Expansion of range for HAS INTERPRETATION

At risk concepts as example.

This is a general issue in the use of ordinal and nominal observables.  The use of DUE TO is suboptimal.  

Concern about defining findings using the INTERPRETS/HAS INTERPRETATION where adjectives are used as the value.  This is a question/answer boundary issue.  The top level concept of finding of at risk is too vague to be useful. There is some ambiguity as to whether low, moderate and high mean absolute or relative to the population.  There is a concern about the meaning of qualifiers other than high risk (i.e. low, moderate, etc.).  What is the context?  Does "increased" risk mean "high" risk?  Does "low" mean lower than the general population or some other cohort? Need to reevaluate the qualifier value hierarchy to ensure consistency in the representation of the relative values.

Some of the risk "due to" are not needed and may be inactivated.

The full model currently being used for "At risk" concepts is defined in the template located at: At [qualifier] risk of [finding/event] (finding) - Ready for review

There was consensus that the Example 2 model was preferable to model 1 (using DUE TO), but there are questions about the meaning of INTERPRETS/HAS INTERPRETATION in the context of risk factors.   The definition of INTERPRETS is "This attribute refers to the entity being evaluated or interpreted, when an evaluation, interpretation, or judgment is intrinsic to the meaning of a concept."  Thus, in the case of 80943009 |Risk factor (observable entity)|, the entity being evaluated is the specific risk factor related to the HAS REALIZATION value (i.e. the disorder or finding being realized).  The definition of HAS INTRPRETATION is "This attribute refers to and designates the judgment aspect being evaluated or interpreted...". In the case of 714664001 |At increased risk of ulcer of foot due to diabetes mellitus (finding)|:

Foot ulcer is the realization

The entity being evaluated (INTERPRETS) is 80943009 |Risk factor (observable entity)| with Diabetes mellitus (judgement) as the value of HAS INTERPRETATION

Need additional examples of where the need for expansion of the HAS INTERPRETATION  would be needed. Jim Campbell will provide a paper describing additional examples and uses for expanding the range.

27 Feb 2023 

A document provided by Monique van Berkum  (Attached above) provides options for resolving the modeling of At risk issue for this topic as well as providing alternative options for consistency in modeling more complex concepts of the pattern: At <risk level> for <Y> due to <Z>.

Proposals include:

• Inactivation of Gravity submitted content of the form "Adverse health risk due to <X>" or "At risk due to <X>" and replacement with findings describing <X>.  This would potentially eliminate the need to expand the range of HAS INTERPRETATION to support risk concepts, although additional testing would be needed in other areas of the terminology (such as survey instruments)
• Creating precoordinated concepts, where needed to represent <Y> due to <Z> in order to be consistent in the use of HAS REALIZATION only vs. either HAS REALIZATION  or both HAS REALIZATION and DUE TO.

This document also underscores some additional issues with functioning concepts that use the INTERPRETS/HAS INTERPRETATION pattern of modeling, especially in the area of negation (or does not).  This is similar to the issue seen in the Situation hierarchy with inverted taxonomies.  While out of scope for this immediate issue, it is something that SNOMED needs to address..

High vs. increased and Low vs. decreased

Similar issue as to the measurement findings where we reassigned values to a reference range.  Initial testing of moving "High (qualifier)" under "Increased (qualifier value)" had no negative impact on taxonomy.  The use of non-specific qualifiers may cause a problem with other areas where "high" is not related to risk. The issue with "decreased" in the area of suicide is that it is unknown where it means "Low" or reduced relative to a prior evaluation.

Increased risk for the most part refers to an increase relative to a "norm" for a population. "High" is absolute related to an individual? Under what circumstances would "High" be used synonymously with "Increased"?  This is dependent on the at risk condition.  Consider using other qualifier values that include "risk"?  Requires additional evaluation. Suggested that we do not accept new "decreased" concepts due to the ambiguity and consider using "decreasing" in the future.  consider inactivating "decreased risk of suicide" as ambiguous. Create "Low risk of suicide".

Expansion of HAS INTERPRETATION continued to the April meeting.

The Content Managers Advisory Group comments indicate that these terms are not or rarely used (probably erroneously) aside from the few concepts in the UK listed above.

Discussion:

Many of these concepts provide some use, but because they cannot be defined they should not be within the SNOMED taxonomy.  For replacement values for high usage, we will determine whether a replacement is needed. Consensus that these concepts be inactivated and a briefing note created for the CoP.

Review of Glossary definition for "Sufficiently defined concept".  See sufficiently defined concept

Additional information on Necessary and sufficient conditions: D.2 Necessary and Sufficient - Examples

Can a concept be sufficiently defined if part of the meaning of the FSN is captured only in the wording of a stated intermediate primitive parent, not in defining attribute-value relationships? For example: Joint laxity (finding) is primitive since it has no defining relationship(s) that capture the "laxity" aspect of the FSN meaning - but can its subtypes (e.g. Elbow joint laxity (finding), Hand joint laxity (finding) etc.) be considered to have a sufficient definition based on the stated parent Joint laxity + their finding site relationship specifying which joint is involved? Currently they are all primitive, but some subtypes of Laxity of ligament are defined on the basis of their intermediate primitive parent and their finding site.

Clear, explicit editorial guidance is needed on this question. The statement here "A concept is sufficiently defined if its defining characteristics are adequate to define it relative to its immediate supertypes" is somewhat ambiguous, since it's not clear whether "defining characteristics" here refers to defining attribute-value relationships specifically or to the logical definition as a whole.

• the phrase "relative to its immediate supertypes" would seem to argue against making a concept defined based in part on information in the FSN of an intermediate primitive parent (since a parent concept, as an immediate supertype itself, obviously cannot contribute to define its own subtype relative to itself. However, that implication may be an artifact of imprecise wording rather than representing actual editorial policy.
  If information represented in the FSN of a stated intermediate primitive parent can be considered to fill a gap in meaning left by the defining attribute relationships, I think that policy should be clearly stated (and ideally illustrated with an example) either here or under Intermediate Primitive Concept Modeling. (Whichever section the guidance is placed in, it might also be helpful to add a "See ____" link to that guidance in the other section as a kind of cross-indexing).

• The SNOMED Glossary states: ""A sufficiently defined concept has at least one sufficient definition that distinguishes it from any concepts or expressions that are neither equivalent to, nor subtypes of, the defined concept". Is "definition status" of a concept part of the definition of a concept? Currently it acts in that way.  This occurs when two or more concepts have the same defining relationships, but only one is marked as sufficient defined.  Those that are not marked as sufficiently defined classify as subtypes.
  • In some cases the SNOMED concept model is inadequate to "fully" define the meaning of an FSN, yet can provide a sufficient definition to make it unique within the terminology.  In these cases the primitive subtypes with the same relationship have the necessary relationships, but not sufficient definitions.
• The Glossary also states "Prior to July 2018, SNOMED CT could only support one sufficient definition for each concept could not represent the 8801005 | Secondary diabetes mellitus (disorder)| example above. A further limitation, that also prevented formal representation of that example was the stated relationship file was only able to represent necessary conditions.". 
  • Proposed clarification: "Prior to July 2018, SNOMED CT could only support one sufficient definition for each concept, and the stated relationships comprising that definition could represent only necessary conditions. A concept such as 8801005 | Secondary diabetes mellitus (disorder)| , which can be caused either by a substance or by another disorder, could not be sufficiently defined within these constraints, since neither | Causative agent| | Substance| nor| Due to| = 64572001 | Disease| is necessarily true. Assigning the concept two sufficient definitions can, however, allow the full meaning of 8801005 | Secondary diabetes mellitus (disorder)| to be formally represented. (See sufficient definition)."
• Lastly, there is a statement: "Following these changes a concept will only be marked as sufficiently defined if it is sufficiently defined by relationships. However, the OWL axioms may provide a sufficient definition that cannot be fully represented as relationships."
  • This has come about with the ability to model concepts with multiple sufficient axioms using GCIs. Proposed clarification: "Following these changes a concept will only be given a definition status of Defined if it is sufficiently defined by the stated relationships in a single axiom. However, multiple OWL axioms may provide a sufficient definition representing different sufficient but not necessary relationships. These concepts will retain the default status Primitive but function as sufficiently defined concepts that will subsume subtypes."

Discussion:

Concepts with identical definitions aside from the Definition status may be classified as supertype/subtypes, due to the limitations of the concept model to allow for more robust definitions. This is being left with the EAG for additional comments to be reviewed at the April meeting. There is a question about whether the browser can be modified to show an icon that can identify a concept as having GCIs.  A question will be forwarded to the tech team.