Occupations in ECTO, including process exposures of activity or occupation?

[laurenechan]

Looking at occupations from EPR and UDN, many of these surveys are asking what occupations individuals have had during a certain period of time or throughout their entire life. Individuals are typically asked to report back their official job title in the survey.

From this, we can likely identify "high risk" occupations like welding, chemical work, etc. But are we interested in modeling
'exposure to X occupation'
OR
'exposure to X activity' (that happens in the occupation)

for example,
'exposure to welding occupation'
OR
'exposure to welding'

For either option we can then add adjunct subclasses such as 'exposure to welding fumes' also. In reality, the questions from the survey are asking about the occupation itself, but the risk/exposure is not coming from the literal job title, it is coming from the work activity. Is there a preferred option as to what we should include for occupations in ECTO?

[matentzn]

My instinct would be to model this more in terms of behaviors (activity), but I think this needs a bit of fleshing out. I think you can do either. I would see this practically? Is there an ontology that models them, so you dont have to curate the exposure to occupation terms from scratch?

[laurenechan]

Currently, terms including occupations or action appear most represented in NCIT and OMIT has terms for health occupations and natural science disciplines. As we have some slightly obscure ones to model, I think we will probably end up adding or requesting terms regardless (think jewelry making and stained glass making).

[matentzn]

I think here you really need to carefully think whether you want ontology terms for these. @dosumis and I have started to recommend, albeit in separate contexts, to consider to model to a useful next nearest term (say manual labour), and then push the rest of the Metadata into a nice blob of searchable text. Do not create ontology terms just because you are faced with a concept; create one only because a concrete use case demands a distinction. Make sure the use case is both required by a user and can be solved by the modelling!

[diatomsRcool]

In this case the use case may require the class. The client wants to be able to reorganize data in new ways. That means the survey may ask about occupation, age of house, and hobbies, but they don't really care about what your job or hobbies are, just that welding exposes you to lead and so does making stained glass art and living in an older house. We should probably talk about this at our next meeting.

[dosumis]

Better to record the actual exposure then - and have the rest as evidence. Unless there's a pressing need to use occupational detail for filtering, consider using free text.

[cmungall]

I agree if there is a need for a named class I would have a DP something like exposure and has-input some (E and during some occupation) label: exposure to %E, occupational this naturally subclasses to the more generic E exposure

…

On Sun, Feb 23, 2020 at 9:04 AM David Osumi-Sutherland < ***@***.***> wrote: Better to record the actual exposure then - and have the rest as evidence. Unless there's a pressing need to use occupational detail for filtering, consider using free text. — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#83?email_source=notifications&email_token=AAAMMOPZUIUQ3LVXHWFPI33REKUAVA5CNFSM4KYWFV32YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEMWBHYA#issuecomment-590091232>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAMMOOPT6BWS5764JVWBITREKUAVANCNFSM4KYWFV3Q> .

[laurenechan]

Sounds like we should plan to model the activity exposure as opposed to the occupation. Next layer to this, for something like exposure to welding fumes, we also have a mixture here in which we need also this to represent exposure to aluminum, antimony etc. https://www.osha.gov/Publications/OSHA_FS-3647_Welding.pdf
Our users want to take their survey data (i.e. someone reports working as a welder etc) and then query it based on who is being exposed to antimony for example.
What is our best approach to construct the exposure to welding class such that we have all the logical relations to these constituent parts?

[matentzn]

If I understand you correctly I think you would like to solve the problem analogous to inheres in vs inheres in part of in the phenotype world. Not sure wether this makes sense to stimuli: Is it really the case that if an event has x as stimulus, it also has all the parts of x as a stimulus? Could be, but this needs to be thought through carefully by @diatomsRcool

exposure to antimony = "'exposure event' and 'has exposure stimulus' some antimony"
exposure to welding fumes = "'exposure event' and 'has exposure stimulus' some 'welding fume'"
antimony subClassOf: part of some 'welding fume'

they wont subclass, but you can now as:
"'exposure event' and 'has exposure stimulus' some part of some 'welding fume'"

and you will get exposure to antimony back. Lets think about it more!

[diatomsRcool]

Not sure wether this makes sense to stimuli: Is it really the case that if an event has x as stimulus, it also has all the parts of x as a stimulus?

This will be true for some things, but not others.