search

phenopackets / phenopacket-format

26 stars 10 forks source link

modeling cohort, no individual data available? #92

Open siiraa opened 7 years ago

siiraa commented 7 years ago

Is there a trick to using phenopacket to model patient's information at population level? E.g. Cohort of 150 colorectal carcinoma patients with hypertension phenotype?

cmungall commented 7 years ago

This is a great example.

No tricks need to be involved. Here the entity would be the cohort. We do not have much documentation on this at the moment. Would the cohort be represented elsewhere with its own URI, or do you need to represent all details of the cohort within the phenopacket?

siiraa commented 7 years ago

Hi, I do not have the definite description of the cohort yet. I'm exploring how best to use phenopacket to model patient's record where I may or may not have the individual's description, but patient's confidentiality will be a factor in capturing some of the data. Given that there will be mulitple complex phenotypes within the same cohort at different frequencies too (that's why lumping the entity to be the cohort entity may be a bit of a work-around? - can the same cohort entity be associated with different phenotypes at a different probability/frequency?).

Documentation on a use case will be great, will be happy to contribute. Does Monarch have a similar scenario?

Thank you.