SNOMED CT terminology data is distributed in release packages containing tab-delimited release files that conform to the SNOMED CT Release Files Specification. All release files follow a consistent design pattern with the following key features:
- A versioning mechanism, which tracks additions, changes, and inactivations made in each release. This versioning mechanism enables terminology services to access the full history of every SNOMED CT component and reference set member.
- A limited range of data types, all of which are rendered as text in release files.
- Use of the Unicode UTF-8 character set for the string data type, to support the inclusion of the full range of Unicode characters.
Data from the files in SNOMED CT release packages needs to be loaded into a datastore that can be accessed by terminology services. The design of this datastore is a key design decision for developers of terminology services since it will determine the capabilities and performance characteristics of those services.
General requirements for the datastore design include:
- It must enable the representation of all the information contained in all release files types in the SNOMED CT release package(s) imported
- It must support the terminology services that are marked as Required in the subsections of 4 Terminology Service Types
- It must deliver scalable optimization of terminology services required by the use cases in 3 Terminology Service Use Cases that the server is designed to support1.
Different technical options are available for implementing terminology services, for example, using a relational database7, other database options (such as Graph databases 4,5,6), or using predefined services accessible via an API (for example SNOMED International's Snowstorm2).
Footnotes
A terminology service provider may choose to limit the range of use cases supported by their server. For example, terminology services required to enable 3.5 Reference Set Editing or 3.8 Support Terminology Authoring and Review may not be supported but some servers. Similarly, some terminology service providers may not optimize services that are used less frequently. For example, terminology services required to enable 3.7 Terminology Change Management are only required when intermittently when terminology update are being reviewed and applied.
Snowstorm is a scalable SNOMED CT terminology server. It can be downloaded and installed from SNOMED International's GitHub site https://github.com/IHTSDO/snowstorm.
Elasticsearch® is a registered trademark of Elasticsearch B.V. (see https://www.elastic.co/elasticsearch/).
Neo4j®, is a registered trademark of Neo4j Inc (see https://neo4j.com/).
A Neo4j upload script for SNOMED CT release files is available on GitHub (see https://github.com/IHTSDO/snomed-database-loader/tree/master/NEO4J). This script was kindly provided by Scott Campbell and his team from the University of Nebraska Medical Center, Omaha,
Papers related to the use of Neo4j with SNOMED CT include: An alternative database approach for management of SNOMED CT and improved patient data queries, Scott Campbell et al, Journal of Biomedical Informatics https://www.researchgate.net/publication/281015927 and Implementation of a Terminology Server with SNOMED CT in Graph Databases, Elizabeth Silva Layes et al, MedInfo 2019 (see http://ebooks.iospress.nl/publication/52311).
The SNOMED CT - SQL Practical Guide includes an example database in which tables are direct matched to the release file structures. The tables are then indexed to optimize searches and joins between the tables that support the delivery of services. The example database is not designed to be used to deliver operational terminology services. However, it provides a more direct illustration of the relationship between the content of the release files and the services outlined in 4 Terminology Service Types. The guide also includes a link to a downloadable script that creates a MySQL database and loads SNOMED CT release files into.
Feedback
Overview
Content Tools
Apps