I couldn’t see a topic for this, but thought I’d kick off the discussion.
I understand the requirement but I don’t think it’s the right path to go down. The requirement can be met with Substances (and even better if the “has biological origin” type property was introduced*).
We have concepts like
933466821000036101|Lactobacillus rhamnosus 10 billion CFU + Saccharomyces boulardii 7.5 billion CFU capsule (clinical drug)| in AMT.
And model these with a substance ingredient.
In fact, 700441006|Saccharomyces boulardii (substance)| is a core substance.
A substance in SNOMED CT is a material entity which is a collective here
They are not molecules or atoms.
They are not objects.
Thus they are also not organisms. They are some unbound quantity of organisms.
That hierarchy is for the names of living organisms.
These medicinal products contain some quantity of these.
The units used for these products “billion CFU” suggest a count (rather than a collective).
But mmol are also a count unit. But we can also measure chemicals in gram.
Just as we can also measure these quantities of microorganisms as gram.
For all intents and purposes these are substances.
Just as a jar of sesame seeds is not a jar of organisms.
Which brings us to the extracts.
”Garlic extract 240 mg capsule” - doesn’t contain “Allium sativum (organism)”.
Even if it wasn’t an extract and each capsule contained a whole clove of garlic - going down the path of modelling parts of organisms is going to lead to a lot more pain than benefit.
If you have a concept 720023002|Dandelion root| in the organism hiearchy. Then you also need to move 256297004|Dandelion pollen| and then what happens with 300910009|Allergy to pollen|
There is an outstanding ticket (many years old) to create a link between substances and organisms that can help with this. Particularly with to allow substances to be unambiguously defined and classify - based on the more unambiguous Linnean taxonomy of the organism hierarchy. (SI Jira is down, I’ll update with links when available).