Context aware terminology binding principles

Information model-driven vs Terminology-driven Context Representation

Overarching Principle

• Ensure that the consolidated meaning represented by the chosen SNOMED CT concept and the applied information model represent the intended semantics, and unambiguously declare the context. I.e. the verification, temporal, and subject relationship context.

Below find a set of terminology binding principles, to help guide the binding of specific types of data elements.

Terminology Binding Principles and Examples

Adhering to these principles provides a robust framework for designing terminology bindings compatible with the SNOMED Concept Model and the structural expressivity of FHIR resources. By ensuring compatibility, we can create a cohesive system where clinical content is accurately and consistently represented across different platforms and applications. This alignment is crucial for maintaining the integrity of clinical data, allowing for seamless integration and reducing the risk of misinterpretation or data loss during exchanges.

Furthermore, this approach facilitates efficient transformations and adaptations in the context of analytics and interoperability. Standardized terminology bindings enable healthcare systems to effectively share and analyze data, supporting advanced analytics, research, and population health management.

Measurable Entities

This section describes how to model clinical data items that capture measurable or observable values. These may include numeric values (e.g., blood pressure) or coded findings (e.g., eye color).

• Concrete values (e.g., numbers with units)

• Coded findings:

• Use Observable Entity for measurable/quantifiable data.

• Use Clinical Finding when answers are selected from a list of predefined, coded options (e.g., eye color).

This section explains how to model whether a clinical finding is present or absent. It includes both direct verification and more complex contextual modeling.

There are two typical approaches:

• Direct Finding: used for current, patient-specific presence/absence

• Contextual Finding: used for historical, familial, or other contextualized interpretations

• Use Direct Finding for simple verification in the current clinical context.

• Use Contextual Finding when temporal, subject, or situational context is required.

• Ensure consistency between meaning binding and value set binding.

• For complex expressions, follow the .

Used when recording whether a condition applies to the patient at the time of data entry.

Used when the finding is not current or applies to someone other than the patient (e.g., family member, historical diagnosis).

• Replace or populate the 408729009 |Finding context| attribute using the chosen qualifier (e.g., “Known absent”).

• Appropriate for modeling:

  • History of conditions

  • Family history

Boolean-style clinical questions require modeling using SNOMED CT concepts with verification qualifiers. These typically involve conditions where only a “yes” or “no” answer is expected.

• Use “Known present” and “Known absent” to model responses.

• If the absence of a finding is implied (e.g., by default), you may choose to record only the “yes” responses.

• Ensure consistent use of Boolean logic across the dataset.

When applying qualifier values to represent the absence of a condition like diabetes mellitus using SNOMED CT an expression can be derived to convey the meaning with explicit context.

Example:

1. Identify the Concept for Diabetes Mellitus: Firstly, we need to identify the SNOMED CT concept representing diabetes mellitus.

   • 73211009 |Diabetes mellitus (disorder)|

2. Representing the Absence/Presence of Diabetes Mellitus: To represent the absence/ of diabetes mellitus, we need to create an expression that explicitly states the absence of the disorder.

This expression clearly indicates that the disorder of diabetes mellitus is absent in this patient at the current point in time or the time specified in the model. It provides a single expression that encapsulates the meaning of the absence of diabetes mellitus with explicit context.

In summary, expressions representing the absence of any specific disorders or conditions in SNOMED CT can be formulated as a SNOMED expression by combining the concept representing the disorder with a qualifier value indicating absence, linked together using the appropriate context model attribute (in this case, 408729009 |Finding context| ). This approach allows for the creation of concise and unambiguously meaningful expressions that convey the absence of any particular condition, all entirely within the SNOMED CT terminology framework.

These expressions must adhere to the SNOMED CT concept model and other constraints to ensure that they extend SNOMED CT in a cohesive and meaningful way, see the