1 of 46

OMOP Standardized Vocabularies. Part 2

Polina Talapova, MD, PhD

OMOP CDM Consultant

Tufts Clinical and Translational Science Institute (CTSI)

1

Health Informatics Elective

Discovering Digital Health Standards

17 August 2023

2 of 46

Content

Recap
Relationships
Hierarchy
Comparing Domain-Centric and Use-Case-Centric Logic in Mappings
Homework

2

3 of 46

Recap

3

4 of 46

Recap

Building the Standardized Vocabularies

= one format + one repository + OHDSI Vocabulary team + community contribution + communication

Access to the Standardized Vocabularies

OHDSI Athena (+ licences for proprietary vocabularies)

Source of Vocabularies: Adopt Versus Build

The majority of standards was adopted. If you want to develop a new dictionary, check for potential existing reference or backbone before creation

Vocabulary Building-Blocks

concepts and their metadata: id, name, synonym, code, domain, class, standardness, validity, period of existence, relationship

Vocabulary Lifecycle:

fields: invalid_reason, valid_start_date, valid_end_date

4

5 of 46

Relationships: Why do we need them?

You may want to understand the relationships in the OMOP CDM to enable more sophisticated data analysis techniques
You may want to comprehend the relationships between concepts in the OMOP CDM to better understand the context of the data you are working with
You may leverage the relationships between concepts in the OMOP CDM to optimize your database queries for mapping.
You may utilize the relationships within the OMOP CDM to develop and maintain Standardized Vocabularies as an external contributor.
You may want to ensure that the new concepts or vocabularies you are adding to align with the existing OMOP relationships.

5

6 of 46

Relationships

predefined semantic connections or associations that indicates how the two concepts are related to each other.
provide additional context about the concepts within the vocabulary.
help to organize and structure the information within the vocabulary.

6

7 of 46

Relationships

Any two concepts in OMOP Vocabulary can have a defined relationship, regardless of their domain or vocabulary.
The nature of the relationship is indicated by a short case-sensitive unique alphanumeric ID in the RELATIONSHIP_ID field of the RELATIONSHIP and CONCEPT_RELATIONSHIP tables.

7

8 of 46

Relationships

Relationships in the OMOP CDM are symmetrical.
If there's a relationship from Concept A to Concept B, there's also an reverse relationship from Concept B to Concept A.
For example, the "Maps to" relationship has a reverse relationship "Mapped from."

8

9 of 46

Relationships

Relationships are stored in the RELATIONSHIP table
Here you can find relevant relationships for your task and their corresponding reverse relationships.

9

10 of 46

Relationships

10

Category	Definition
Attribute Relationships	Describe attributes or properties associated with concepts, providing additional context or characteristics.
Mapping Relationships	Establish links between concepts from different vocabularies, allowing cross-referencing and interoperability.
Hierarchical Relationships	Define parent-child relationships between concepts, creating a hierarchy for navigation and organization.
Cross-vocabulary Relationships	Connect concepts from separate vocabularies, enabling integration and comparison across diverse sources.
Relationships Between Concepts of the Same Vocabulary	Establish relationships between concepts within a single vocabulary, aiding in grouping and classification.

11 of 46

Attribute relationships: LOINC

11

https://loinc.org/search/

2340-8 Glucose [Mass/volume] in Blood by Automated test strip

12 of 46

Attribute relationships: LOINC

12

https://athena.ohdsi.org/search-terms/terms/3011424

13 of 46

Attribute relationships: SNOMED

13

https://browser.ihtsdotools.org/?perspective=full&conceptId1=404684003&edition=MAIN/2023-07-31&release=&languages=en

SCTID: 52734007 Total replacement of hip

14 of 46

Attribute relationships: SNOMED

14

https://athena.ohdsi.org/search-terms/terms/4203771

15 of 46

Attribute relationships: RxNorm

15

https://mor.nlm.nih.gov/RxNav/#

16 of 46

Attribute relationships: RxNorm

16

https://www.ohdsi.org/web/wiki/doku.php?id=documentation:vocabulary:drug

17 of 46

Attribute relationships: RxNorm

17

https://athena.ohdsi.org/search-terms/terms/40228229

Drug Products

18 of 46

Attribute relationships: RxNorm

18

https://athena.ohdsi.org/search-terms/terms/40228229

19 of 46

Mapping relationships: Maps to (1)

19

20 of 46

Mapping relationships: Maps to (2)

20

21 of 46

Mapping relationships: Maps to (3)

21

22 of 46

Mapping relationships: �Maps to + Maps to value (1)

22

23 of 46

Mapping relationships:

Maps to + Maps to value (2)

23

24 of 46

Mapping relationships:

Maps to + Maps to value (3)

24

https://ohdsi.github.io/CommonDataModel/vocabulary.html

25 of 46

Mapping relationships:

Maps to + Maps to value (2)

25

26 of 46

Mapping Metadata in OMOP

concept_relationship.valid_start_date
concept_relationship.valid_end_date
concept_relationship.invalid_reason�

26

27 of 46

SSSOM Operational Model

27

Mapping Metadata In An Ideal World

28 of 46

The MAPPING _METADATA table

CREATE TABLE mapping_metadata (

subject_id VARCHAR,

predicate_id VARCHAR,

object_id VARCHAR,

mapping_source VARCHAR,

mapping_set_id VARCHAR,

mapping_set_label VARCHAR,

mapping_provider VARCHAR,

mapping_date DATE,

author_id VARCHAR,

author_label VARCHAR,

reviewer_id VARCHAR,

reviewer_label VARCHAR,

mapping_tool VARCHAR

mapping_tool_version VARCHAR,

confidence FLOAT,

mapping_justification VARCHAR,

subject_source_version VARCHAR,

subject_category VARCHAR

);

28

Mapping Metadata In An Ideal World

29 of 46

Cross-Vocabulary Relationships

SNOMED - LOINC https://athena.ohdsi.org/search-terms/terms/4144235

29

30 of 46

Cross-Vocabulary Relationships

ATC - RxNorm

https://athena.ohdsi.org/search-terms/terms/21604709

30

31 of 46

Internal Vocabulary Relationships

LOINC - LOINC

https://athena.ohdsi.org/search-terms/terms/1002319

31

32 of 46

Internal Vocabulary Relationships

SNOMED - SNOMED

https://athena.ohdsi.org/search-terms/terms/4149320

https://athena.ohdsi.org/search-terms/terms/44790677

32

33 of 46

Hierarchy

Within a domain, standard and classification concepts are hierarchically organized and stored in the CONCEPT_ANCESTOR table.
This table enables querying and retrieval of concepts and their hierarchical descendants.
Descendants inherit attributes from their ancestors and may have additional or more defined attributes.

33

34 of 46

Hierarchy

CONCEPT_ANCESTOR table is automatically built from the CONCEPT_RELATIONSHIP table, traversing all possible concepts connected through hierarchical relationships.
You can check whether relationship_id is hierarchical or not using the is_hierarchical field in the RELATIONSHIP table

34

35 of 46

Hierarchy

Hierarchical relationships include "Is a" - "Subsumes" pairs and other relationships connecting hierarchies across vocabularies.
Participation in the hierarchy constructor is determined by the flag DEFINES_ANCESTRY in the RELATIONSHIP table for each relationship ID.

35

36 of 46

Hierarchy

Ancestral degree, capturing the number of steps between ancestor and descendant, is indicated by the MIN_LEVELS_OF_SEPARATION and MAX_LEVELS_OF_SEPARATION fields.

36

37 of 46

Hierarchy

Not all hierarchical relationships equally contribute to levels-of-separation calculation.
The step counted for the degree is determined by the IS_HIERARCHICAL flag in the RELATIONSHIP table for each relationship ID.

37

38 of 46

Hierarchy

Comprehensive hierarchies in OMOP exist for the Drug and Condition domains.
Procedure, Measurement, and Observation domains are partially covered and still in progress.
Community contributions are warmly welcomed.

38

39 of 46

Hierarchy: Condition

39

40 of 46

Hierarchy: Drug

40

41 of 46

Domain-Centric vs. Use-Case-Centric Logic in Mappings

41

Feature	Domain-Centric Logic	Use-Case-Centric Logic
Definition	Mapping logic is guided by the known domains of the source tables (e.g. procedures, lab tests)	Mapping logic is driven by a specific use case, like mapping psychiatric assessment tools, which may not be tied to one specific domain.
Scope	The mapping is broad and encompasses a wide range of entities from a source table	The mapping is focused on a specific use case and includes only the entities related to that use case.
Flexibility	Less flexible as it's tied to a specific domain.	More flexible as it can be tailored and data can be stored in different domains.
Complexity	Relatively simple as it follows predefined domain rules.	Can be more complex as it needs to handle various source types and different rules depending on the use case.

42 of 46

Domain-Centric vs. Use-Case-Centric Logic in Mappings

42

Feature	Domain-Centric Logic	Use-Case-Centric Logic
Mapping Rules	Predetermined by the domain.	Customized based on the source data and use case
Data Granularity	Tends to have a fixed level of granularity as it follows domain rules.	Can have varying levels of granularity depending on the use case and source data.
Value Storage	Follows domain conventions, e.g. Measurements and Observations can use value_as_concept_id and value_as_number; Procedures can't.	Flexibility to store values depends on the use case.

43 of 46

Domain-Centric vs. Use-Case-Centric Logic in Mappings

43

Feature	Domain-Centric Logic	Use-Case-Centric Logic
Clinical Data Tables	Tables are populated based on domain-specific rules.	Tables are populated based on the use case, focusing on attributes relevant to the specific assessment tools.
Data Quality	Data quality assurance follows domain-specific conventions, with validation based on domain rules.	Data quality assurance is tailored to the use case, with validation focused on the specific attributes and requirements.

Both scenarios can uncover gaps in the OMOP Vocabulary at both the domain level and the use case-based vocabulary level.
In future sessions, we will discuss how to address these gaps by adding your own content to the OMOP Vocabulary.

44 of 46

OHDSI Tools using OMOP Standardized Vocabularies

44

45 of 46

Homework

To develop a small synthetic vocabulary (10-15 concepts) with internal and/or external hierarchies.
The content should be stored in the concept_manual, concept_synonym_manual, concept_relationship_manual, and vocabulary tables, as shown in the OHDSI Template #4 for adding a new vocabulary.
Please send it to polina.talapova@tuftsmedicine.org when ready.

45

46 of 46

Questions?

46