SNOMED CT: An Introduction — Part 2

SNOMED CT: An Introduction — Part 2

Continuing from the SNOMED CT: An Introduction — Part 1 article, let’s now talk about how SNOMED CT represents meaning of medical ideas, how it facilitates reuse of data and its wider role in interoperablity.

Meaning of Medical Ideas (concepts) in SNOMED CT

As we have said SNOMED CT covers multiple care settings. One way of catering for all the different care settings is that SNOMED CT supports the difference in language used by clinicians, e.g. one clinician might say heart attack, another cardiac infarction and another myocardial infarction, but they would all come under the same concept 22298006 |Myocardial infarction (disorder)| with the synonyms allowing for the different ways of saying the same thing. Synonyms allow for different expressions with the same meaning.

In addition to synonyms, there is the meaning of medical ideas (concepts) represented in SNOMED CT. For example, this allows us to use SNOMED CT to determine that 882784691000119100 |Pneumonia caused by SARS-CoV-2| is caused by 840533007 |Severe acute respiratory syndrome coronavirus 2 (organism)|as shown in the diagram below.

It is this meaning that is represented in SNOMED CT that allows machines (and humans) to reuse SNOMED CT in more powerful ways. From the above example, we could infer that this 882784691000119100 |Pneumonia caused by SARS-CoV-2|

is also a type of:

  • 840539006 |Disease caused by severe acute respiratory syndrome coronavirus 2
  • 398447004 |Severe acute respiratory syndrome|

In fact, SNOMED CT comes with this information built and such inferences are called `Parent` relationships as show below.

Reusing Medical Data with the power of SNOMED CT

The best part of SNOMED CT is that such inferences are automatically provided and more importantly, these logical inferences can be used by machines at runtime to do clever stuff.

For example, imagine a case where we would want to count the number of Viral infections that were treated in the last 12 months, we would want to count all cases of 882784691000119100 |Pneumonia caused by SARS-CoV-2| in them.

Now imagine a different research study wants to collect all patients with heart disease who also had a `Respiratory disease` in the last 6 months. Clearly, the researcher would want to include all cases with 882784691000119100 |Pneumonia caused by SARS-CoV-2| in their study.

Now based on the meaning of `​​882784691000119100 |Pneumonia caused by SARS-CoV-2|` built into ​SNOMED CT, we were able to identify these cases for both studies which at the surface seem to be looking for very different things — Viral Infections and Respiratory Infections! This unlocks the ability to collect data once and reuse it for multiple purposes.

That is the true power of the `meaning` in SNOMED CT — the secret power of the `Relationships` and the ability to use them dynamically. The best part of this, is that using SNOMED CT’s rules, we can allow machines (algorithms, AI) to create such inferences on real world data!

SNOMED CT and Interoperablility

If the same structured terminology as SNOMED CT is used across different care settings, then the information transferred between these care settings would not require converting and would, therefore, lose none of its meaning. This transfer of information between systems is known as interoperability.

If the clinical information is recorded using the same structured terminology then it can be transferred between the systems losing none of its meaning as there is no need to ‘convert’ it.

In this example, a patient visits their GP, makes an appointment to see a consultant via the NHS e-Referral Service (e-RS), is treated by the consultant and the relevant discharge summary is sent to the GP.

Now imagine a situation where the receiving system has a rule that says all infectious disease referrals should be automatically redirected to their Specialist Hospital. Normal referral systems would have to write additional rules to do such redirection — they would need to know a list of all infectious diseases (which is tens of thousands!! 😂). However, a receiving system capable of truly using SNOMED CT can infer that any condition for example 882784691000119100 |Pneumonia caused by SARS-CoV-2| is an Infectious Disease and automatically redirect the patient.

In fact, any referral system that can tap into the power of SNOMED CT relationships would be able to automatically filter the choice of referral centres based on the condition selected.

SNOMED CT & Content Updates

Clinical medicine is constantly evolving and new diseases and tests are constantly being introduced based on research — MonkeyPox being a more recent example, after CoVID-19. SNOMED CT is a dynamic product, constantly being updated and, at the moment. This means that the SNOMED CT UK Edition is updated every six months. These content updates also mean that the inferences themselves can be dynamically updated based on the updated meaning of a concept in SNOMED CT. Of course, some of you might be thinking — hang on, how we can control what changes are made to the inferences! You’d be on the money to want to see these inferences before they hit any system you might have built to use these inferences.

The good news is that there are some well understood mechanisms for tracking these changes and for processing these updates to SNOMED CT. These will likely be an upcoming article, so watch the space!

Want to discuss more about SNOMED CT or Interoperability?

We hope you found this second part of our SNOMED CT introduction series interesting. If you have any questions or feedback, please shoot us an email at [email protected].

Want to read more? Check out the first and third parts of this article below!
Part 1 – https://termlex.com/snomed-ct-an-introduction-part-1/
Part 3 – https://termlex.com/why-are-snomed-ct-codes-identifiers-meaningless-numbers-snomed-ct-intro-part-3/

Check us out on Medium, LinkedIn, and Twitter!

Other resources

Categories: Ontology, SNOMED CT, Standards

Tags: ,

Newsletter