This page describes the grouping of attributes.

A relationship group combines an attribute-value pair with none, one, or multiple attribute-value pairs in order to refine the meaning of a concept.  

For example,

Stated view of 18876004 |Pain in finger (finding)| with the Finding site (attribute) and its value of Finger structure (body structure)

An attribute must be populated with a target value to model a concept. 

Relationship groups are needed when modelling:

• • Clinical finding concepts that require multiple Associated morphology attributes and multiple Finding site attributes
  • Procedure concepts that require multiple Method attributes and multiple Procedure site attributes.

• A single relationship group containing only one attribute can exist.
  • When an attribute is restricted to a single group with no other attributes, the attribute is described as being "self-grouped".

• Multiple attributes may be grouped together in relationship groups, and multiple relationship groups may be created to sufficiently define concepts. 
• When creating new concepts or revising existing ones, each attribute type included in a relationship group may only be present once, e.g. two Associated morphology attributes cannot be in the same relationship group.
• Relationship groups are not limited to Clinical finding and Procedure concepts.
• There is no limit to the number of relationship groups that may be added to a concept.

Ungrouped attributes

An attribute that is not in a relationship group is considered to be in a group on its own. When attributes are not grouped, their meanings are interpreted separately. For example, in the following diagram, the Associated morphology is Hemorrhage, and the Finding site is Uterine structure. However, it cannot be interpreted that the site of the Hemorrhage is the Uterine structure because the two attributes are not grouped. 

Inferred view of self-grouped attributes values of Hemorrhage (morphologic abnormality) and Uterine structure (body structure)

When the attributes are grouped, the relationships imply meaning towards each other.  To continue the example above for  , the following diagram shows the Associated morphology of Hemorrhage and the Finding site of Uterine structure in a relationship group together.  The grouping can be interpreted that the finding site of the hemorrhage is the uterine structure. 

Inferred view of grouped attribute values of Hemorrhage (morphologic abnormality) and Uterine structure (body structure)

Note the difference in the inferred parents between the self-grouped versus grouped attributes. This is explained in more detail below.  

Impact of relationship grouping on inheritance

Relationship groups refine inheritance, i.e. a grouped set of attributes is more specific than the same attributes that are not grouped. This is important when considering subsumption. The following diagrams demonstrate the impact of grouping or failing to group consistently using the concepts  and one of its supertypes, . 

The meaning of the supertype concept,  (where the relationships are grouped) is interpreted as a procedure with an excision on the aortic structure. This is because  and are grouped. 

Inferred view of Excision of aorta (procedure) with grouping of attributes

In the following diagram, the more general supertype concepts,  and  are the proximal supertype concepts.

 is a logical subtype of .  However, the attributes of the concept  are not grouped. Thus, the classifier interprets the definitions as non-related and  is not inferred as a subtype of . This is because the attributes in the subtype concept are not grouped, i.e are not explicitly stated. From a machine-processing perspective, each attribute is considered a group on its own; i.e. there is an excision, but nothing else is known about the excision. This results in the concept,  , being interpreted more broadly. 

 Inferred view of Excision of lesion of aorta (procedure) without grouping of attributes

In the following diagram the attributes of the concept   are grouped. An author that explicitly states that the excision is of a lesion found in the aortic structure, by grouping the attribute-value pairs, provides the necessary information for the classifier. This enables   to be inferred as a subtype of  .  

Inferred view of Excision of lesion of aorta (procedure) with grouping of attributes

Same attributes in separate relationship groups

Each relationship group should only contain one instance of an attribute. This is because two of the same attributes in a relationship group is not the same as one attribute with one target value that captures the combined meaning of the target values, as illustrated in the following diagram.

Two Finding site attributes are required to support the location of . Each  and its respective target value are placed in a relationship group with the attribute  with its target value of .  

Inferred view of Associated morphology (attribute) with its value of Fracture, closed (morphologic abnormality) in two separate relationship groups

Procedure hierarchy

In the hierarchy, a relationship group is usually a way of combining attributes about a particular method. 

In the concept   within the following diagram, the relationship groups clarify that there is exploration of the bile duct and excision of the gallbladder. Without the relationship groups, the appropriate relationships between the attributes would be unclear; i.e. the exploration of the bile duct versus gallbladder and the excision of the bile duct versus the gallbladder.

Inferred view of a Procedure hierarchy relationship group: combining attributes around Method (attribute)

Clinical Finding/Disorder hierarchy

In the Clinical finding/Disorder hierarchy:

Stated view of a disorder hierarchy concept with Causative agent and Pathological process attributes in the same relationship group

Observable Entity hierarchy

When defining concepts, attributes are self-grouped.  Each observable entity represents only one property being observed.

For example,  is represented using multiple attributes with each in its own relationship group.

Stated view of a concept from the Observable entity hierarchy with self-grouped attributes