CIMI Terminology Binding Dr Linda Bird 13 th

CIMI Terminology Binding Dr Linda Bird 13 th April 2013

Agenda • • • Use Cases and Requirements Proposed Approach Example Lab Results Bindings Terminology Reference Sets Archetype Object Model Support Future Work

USE CASES AND REQUIREMENTS

Use Cases for Terminology in Models 1. Management and quality control of model libraries a) b) Searching model libraries (e. g. Find all archetypes with a meaning << Observable Entity) Identifying semantic overlap between models (e. g. 2 models that contain a cluster c) Inconsistency of model interdependencies (e. g. the meaning of a constrained archetype whose elements have the same or similar meanings) is not subsumed by the meaning of the base archetype) 2. Transforming between isosemantic representations of the model: both a) b) Different levels of precoordination (e. g. ‘Left leg’ versus ‘Leg’ + Laterality= “Left’) Different representation models (e. g. All ENTRYs with meaning << |Observable entity| to be mapped to a HL 7 v 3 Observation) 3. Querying data instances of models (including clinical decision support) which use different representations – for example: a) Different level of precoordiation versus structure (e. g. ‘Left leg’ versus ‘Leg’ + b) Different modeling design choices (e. g. Representing a medication’s Indication as a data c) Subsumption testing of values Laterality= “Left’) element, versus an ‘indication’ link to a Diagnosis archetype) 4. Supporting data validation and semantic interoperability (e. g. Exchanging data between systems which use different native information structures)

Requirements for using Terminology in Models 1. Standard (reproducible) way of doing terminology bindings 2. The ability to represent the valid set of values for a given coded element. 3. The ability to state the association between the intended interpretation of nodes in the model and concepts in the terminology 4. Terminology bindings that are agnostic as to whether nodes are connected using a hierarchy or using links. 5. Terminology bindings that allow the values to be represented in a way that is agnostic to the degree of precoordination versus structure. 6. Terminology bindings that enable the transformation between isosemantic representations of the same model 7. Terminology bindings that allow consistency to be checked within models, and between models related by specialisation or used to fill slots (using an underlying ontology).

Management and Quality Control of Model Libraries Example Scenarios • Search for: – An archetype whose meaning ( context) is subsumed by ‘Cardiovascular Observable’. • Validate archetype specialisations: – To ensure that there is a valid relationship between the meanings of the base and the constrained archetypes – for example: o Not valid: ‘Pulse rate’ (meaning = |pulse finding|) based on ‘Heart rate’ (meaning = |heart rate|) Meaning from different hierarchies o Valid: ‘Pulse rate’ (meaning = |pulse|) based on ‘Heart rate’ (meaning = |heart rate|) Meaning of constrained archetype subsumed by meaning of base archetype o Valid: ‘Family history of diagnosis’ based on ‘Diagnosis’ archetype • Validate archetype slot fillers: – To ensure that the meaning of the slot and the meaning of the archetype that fills it are consistent – for example: o Valid: Using a ‘Problem diagnosis’ archetype (meaning = |clinical finding|) to fill a ‘Cardiovascular problem/diagnosis’ slot (meaning = |cardiovascular finding|). o Using a ‘Problem diagnosis’ archetype to fill the following slots in a discharge summary: Ø ‘Family history’, ‘Past history’, ‘Current problem/diagnosis’, or ‘Problems’ o Using a ‘Medication’ archetype to fill the following slots in a discharge summary : Ø ‘Ceased medication’, ‘Current medication’, or ‘Past Medication’

PROPOSED APPROACH

Terminology Binding Approach • The meaning of each node is separated into 3 parts: – Relationship: The relationship from the parent node to this node – Object: The ‘class’ of things defined by this node’s values – Modifier: The context of the node’s meaning – including Subject-relationship, temporal, procedure/finding context, negation, state, certainty • Note: ‘Subject’ of ‘Subject-Relationship-Object’ triple is the parent node

CIMI Terminology Binding Approach STRUCTURE BINDING TERMINOLOGY Meaning Cluster: Medication Value Set Relationship Object Modifier (Linkage concept) Pharm/biol product (Context values) - Element: Medication Name (Linkage concept) Pharm/biol product (Context values) Medication Ref_Set Element: Active ingredient Has active ingredient Substance (Context values) Substance Ref_Set Element: Basis of Strength Has basis of strength substance Substance (Context values) Substance Ref_Set Element: Strength Has strength Measuremen t Finding (Context values) - Element: Dose form Has dose form Drug dose form (Context values) Dose_Form Ref_Set Indication Has indication Clinical Finding (Context values) Indication Ref_Set Element:

Specialising Archetype Meaning (Object) STRUCTURE BINDING TERMINOLOGY Meaning Cluster: Oral Medication Value Set Relationship Object Modifier (Linkage concept) Oral dosage form product (Context values) - Element: Medication Name (Linkage concept) Oral dosage form product (Context values) Oral Medict Ref_Set Element: Active ingredient Has active ingredient Substance (Context values) Substance Ref_Set Element: Basis of Strength Has basis of strength substance Substance (Context values) Substance Ref_Set Element: Strength Has strength Measuremen t Finding (Context values) - Element: Dose form Has dose form Oral dosage form (Context values) Oral Dose_Form Ref_Set Element: Indication Has indication Clinical Finding (Context values) Indication Ref_Set

Specialising Archetype Meaning (Relationship) STRUCTURE BINDING TERMINOLOGY Meaning Cluster: Medication with Primary Indication Value Set Relationship Object Modifier (Linkage concept) Pharm/biol product (Context values) - Element: Medication Name (Linkage concept) Pharm/biol product (Context values) Medication Ref_Set Element: Active ingredient Has active ingredient Substance (Context values) Substance Ref_Set Element: Basis of Strength Has basis of strength substance Substance (Context values) Substance Ref_Set Element: Strength Has strength Measurem ent Finding (Context values) - Element: Dose form Has dose form Drug dose form (Context values) Dose_Form Ref_Set Indication Has primary indication Clinical Finding (Context values) Indication Ref_Set Element:

Specialising Archetype Meaning (Modifier) STRUCTURE BINDING TERMINOLOGY Meaning Cluster: Current Medication Value Set Relationship Object Modifier (Linkage concept) Pharm/biol product Current - Element: Medication Name (Linkage concept) Pharm/biol product (Context values) Medication Ref_Set Element: Active ingredient Has active ingredient Substance (Context values) Substance Ref_Set Element: Basis of Strength Has basis of strength substance Substance (Context values) Substance Ref_Set Element: Strength Has strength Measuremen t Finding (Context values) - Element: Dose form Has dose form Drug dose form (Context values) Dose_Form Ref_Set Indication Has indication Clinical Finding (Context values) Indication Ref_Set Element:

Filling Archetype Slots STRUCTURE Composition Element: Cluster: BINDING TERMINOLOGY Discharge Summary Medical record number Primary diagnosis Diagnosis Element: Diagnosis name Element: Onset datetime Meaning Relationship Object Modifier Has primary diagnosis Clinical Finding (Context values) Meaning Relationship Object Modifier Has diagnosis Clinical Finding (Context values) Value Set -

Filling Archetype Slots STRUCTURE Composition Element: Cluster: BINDING TERMINOLOGY Discharge Summary Medical record number Family history Meaning Relationship Object Modifier Has primary diagnosis Clinical finding Family member Meaning Diagnosis Element: Diagnosis name Element: Onset datetime Relationship Object Modifier Has diagnosis Clinical finding (Context values) Value Set -

EXAMPLE LABORATORY RESULTS MODEL BINDINGS

Laboratory Test Request Summary ENTRY constrains Clinical Entry constrains Clinical Activity constrains Request constrains Observation Request constrains Laboratory Test Request Summary

Clinical Entry

Clinical Entry & Clinical Activity constrains

Clinical Activity & Request constrains

Request & Observation Request constrains

Observation Request & Laboratory Test Request Summary constrains

TERMINOLOGY REFERENCE SETS

Categories of value sets 1. Clinical value sets • For these we will try to always use SNOMED CT, with the addition of CIMI extension concepts where required. 2. Non-clinical value sets, with a single authoritative ‘source of truth’ (e. g. IANA media types, country codes) • For these we will take a copy of the value set into our terminology server, so that the values are available during the authoring process and instance generation. 3. Non-clinical value sets, with no single authoritative ‘source of truth’ (e. g. participation mode) • For these we will provide a maximal set of terms that provides coverage of all member’s value sets, and include a hierarchy that indicates the relationship between a value and its specialisations.

General Principle Value sets which may either be represented in the structure or precoordinated in the definition of another clinical concept (e. g. ‘units of measure’ may be used to define the strength of a medication) would be represented using SNOMED CT, to ensure that the concept definitions can be incorporated into SNOMED CT for isosemanticity.

CIMI Reference Sets Id Name Terminology Model attribute/element 1 2 CIMI_action_type_refset CIMI_composition_category_refset SNOMED CT CIMI value set (non-clinical) action_type COMPOSITION. category 3 CIMI_finding_status_refset SNOMED CT 4 5 CIMI_formalism_refset CIMI_identifier_type_refset 6 CIMI_language_refset CIMI value set (non-clinical) SNOMED CT ISO-639 -1/2 [+ ISO-3166 -1] (e. g. 'en', 'en-GB', 'en-US') CIMI-CLUSTER. action CIMI-RM CIMICLUSTER. observation_result_item CIMI-RM 7 8 9 CIMI_link_meaning_refset CIMI_link_type_refset CIMI_mapping_purpose_refset 10 finding_status CIMI-RM CIMI_media_type_refset SNOMED CT IANA media types (e. g. 'html', 'csv', 'jpeg') PARSABLE. formalism IDENTIFIER. type COMPOSITION. language ENTRY. language TEXT. language LINK. meaning LINK. type TERM_MAPPING. purpose CIMI-RM MULTIMEDIA. media_type 11 CIMI_null_flavour_refset HL 7 v 3 Null flavors CIMI-RM ELEMENT. null_flavor 12 CIMI_observation_interpretation_category _refset SNOMED CT 13 CIMI_observation_result_refset SNOMED CT 14 15 16 17 CIMI_observation_result_group_refset CIMI_participation_function_refset CIMI_participation_mode_refset CIMI_party_relationship_refset SNOMED CT CIMI value set (non-clinical) SNOMED CT 18 CIMI_procedure_status_refset SNOMED CT 19 20 21 22 23 24 CIMI_proportion_type_refset CIMI_structure_type_refset CIMI_symbol_refset CIMI_territory_refset CIMI_units_duration_refset CIMI value set (non-clinical) SNOMED CT CIMI-RM CIMI-ENTRY. observation CIMICLUSTER. observation_result_item CIMI-ENTRY. observation CIMI-RM CIMI-CLUSTER. action_information CIMI-ENTRY. procedure CIMI-ENTRY. observation CIMI-RM CIMI-RM results/interpretation_category name results/result_group_name PARTICIPATION. function PARTICIPATION. mode PARTY_RELATED. relationship action/action_status procedure/status observable/status PROPORTION. type CLUSTER. structure_type ORDINAL. symbol COMPOSITION. territory QUANTITY. units DURATION. units

CIMI_link_meaning_refset

ARCHETYPE OBJECT MODEL SUPPORT

AOM 1. 5 Ontology

Option 1 – Make binding triple explicit To define the ‘relationship-object-modifier’ triplet as an allowable binding statement. ontology term_bindings = < ["/data[cimi-CLUSTER. observe_action] "] = < relationship = <[SNOMED_CT: : 5635636|Has related action|]> object = <[SNOMED_CT: : 123456|Observation procedure|]> modifier = <[SNOMED_CT: : 288529006 |Context values|]> ["/data[cimi-CLUSTER. report_action] "] = < relationship = <[SNOMED_CT: : 5635636|Has related action|]> object = <[SNOMED_CT: : 243256|Report procedure|]> modifier = <[SNOMED_CT: : 288529006 |Context values|]> > >

Option 2 – Use SNOMED CT CG Use SNOMED CT Compositional Grammar inline. ontology term_bindings = < ["/data[cimi-CLUSTER. observe_action]"] = <[SNOMEDCT: : {5635636|Has related action|=123456|Observation procedure|: 288529006|Context values|]>

FUTURE WORK

Future Work • Value bindings: – List the full set of reference sets required – Populate these reference sets • Semantic bindings: – Complete semantic bindings for Model Patterns – Complete semantic bindings for Lab Results Models • Explore the relationship between the ‘Modifier’ binding and data elements, such as ‘Status’, ‘Certainty’, and ‘Negation flag’. • Define other rules and principles • Complete Terminology Binding Style Guide

QUESTIONS