SNOMED CT stands for Systematized Medical Nomenclature for Medicine–Clinical Terminology and is a clinical terminology system that provides a standardized and scientifically validated way of representing clinical information captured by clinicians.

SNOMED CT is a special type of medical coding system called terminology system where the meaning of a term is coded instead of the term itself. It is a controlled medical vocabulary containing standard terms and that codes the concept definition (meaning) of a term. This definition is a formal logical representation of the information that is always necessarily true about its meaning.

This information representing the meaning is modelled as a set of interrelated triples, which encodes it in a structured format thereby making it machine-interpretable. Being always necessarily true under all circumstances, the meaning is interpreted exactly the same way everywhere each and every time.

The information model is made up of the three entities: subject, predicate and object (hence called semantic triples) that are represent the concept’s relationships.

The SNOMED concept model mandates that at least one of the triples must have the concept as the subject, the relationship “is a” as the predicate, and another concept as the object which represents a broader definition.

This makes the subject concept a finer representation or subtype of the object concept. There can be any number of additional triples that are either subtype relationships or attribute relationships, which have different relationship types as predicates.

Example: Appendicitis is coded by its concept definition. Note the three distinct components of each line – together they form the “triple”.

1. “Appendicitis” “Is a” “Disorder of appendix”

2. “Appendicitis” “Is a” “Inflammation of large intestine”

3. “Appendicitis” “Finding site” “Appendix”

4. “Appendicitis” “Associated morphology” “Inflammatory morphology”

This makes the concept of “Appendicitis” to be a finer version of the “Disorder of appendix” concept (there can be other subtype finer concepts of this like “Appendicular abscess”, “Intussusception of appendix”, “Appendicular neoplasm”, etc.) and of “Inflammation of large intestine” concept (whose other subtype finer concepts would be “Colitis”, “Crohn’s disease”, “Proctitis”, etc.). Incidentally, “Appendicitis” concept has several subtype finer concepts such as “Acute appendicitis”, “Chronic appendicitis”, “Atypical appendicitis”, and more.

As is clearly evident from above, all of the four assertions are always necessarily true of the meaning of “Appendicitis” irrespective of any other consideration.

Combining the above, as per SNOMED CT Expression Constraint Language (ECL), the unique language used to express the concept definition is as follows:

=== 18526009 |Disorder of appendix (disorder)| + 302168000 |Inflammation of large intestine (disorder)| : { 363698007 |Finding site (attribute)| = 66754008 |Appendix structure (body structure)|, 116676008 |Associated morphology (attribute)| = 409774005 |Inflammatory morphology (morphologic abnormality)| }

From the above it is evident that the individual components of the triples are in codes making the entire definition machine-interpretable.

Every concept definition is assigned a unique machine-processable identifier.

Example: Code for Appendicitis is 74400008.

Every concept is assigned a unique human-readable official name that explains the concept in a non-vague manner, which is called Fully Specified Name, or FSN, that has the concept’s domain enclosed in parentheses as a suffix to help disambiguate it.

Example: FSN for Appendicitis is Appendicitis (disorder).

All terms synonymous with a concept’s definition is assigned the same concept code.

Example: Inflamación aguda del apéndice [Spanish], Appendicite [French], blindedarmontsteking [Dutch], Appendicit [Swedish] – all have the same code 74400008.

SNOMED CT also permits the coding of terms without matching concept definitions in a standardised manner, post-facto, for machine-processability and machine-interpretability of user specific or local terms.

The concepts can be constrained, extended, or mapped to other code systems as required. All of the resultant codes will be standards-compliant, making the code system very flexible and customisable.

Example:

1. Myalgia relieved by analgesic === 706955002 |Pain relieved by analgesic (finding)| : 363698007 |Finding site (attribute)| = 127954009 |Skeletal muscle structure (body structure)|

2. operable tumour === 128303001 |Surgical removal (procedure)| : 363700003 |Direct morphology| = 108369006 |Neoplasm (morphologic abnormality)|

3. GC fair === 162669003 |Patient’s condition satisfactory (finding)|

The relationships of the concept definition can be used for in-depth query and its machine-interpretability makes all SNOMED coded information Artificial Intelligence (AI) and Machine Learning (ML) ready – imparts “intelligence”.

Example: Appendicitis has many “is a subtypes” of its own and it itself “is a subtype” of others. These relationships as well as its attribute relationships (assertions # 3 & 4 in the above example) can be used to do semantic search or to generate custom cohorts on-demand or trigger automated follow on action using preset rules like “for all concepts having inflammatory morphology as associated morphology prescribe antibiotics”. Since the code is sufficient for this, on user entry, the required action can be performed by the machine without any further action by the user.

In conclusion, SNOMED CT can be said to be a dictionary of healthcare for machines to “understand” human understandable terms.