OHDSI Home | Forums | Wiki | Github

Atlas prediction: what is included in the covariates settings table?


(panpan) #1

dear all,
I use the atlas to do prediction and in the analysis settings and covariates settings, for example, if I choose the domain of observation like the picture. The first question is I am wondering the included covariates are only the observation_concept_id, or they also include observation_ type_concept_id and value_as_concept_id ,qualifier_concept_id,unit_concept_id and so on. how to explain it?

thank you very much

thank you very much


(George Argyriou) #2

Measurement works as a categorical variable. So it checks if patients had e.g. a HDL measurement or not.
Measurement - Value works as a continuous variable. It looks at the value of the measurement e.g. HDL:83.

In Characterization, you’ll get different statistics. From the book of OHDSI:

While domain covariates are computed using a binary indicator (i.e. was a record of the code present in the prior timeframe), some variables provide a continuous value such as the age of persons at cohort start.


(panpan) #3

thank you very much @George_Argyriou, so covariates in the measurement are category covariates and measurement_value are continuous covariates in the model. is it right?
And it means if I choose the measurement, just like the red box in the picture of the bookofohdsi, it just contains the observation_concept_id,right?

thank you very much


(George Argyriou) #4

Yes

And it means if I Choose the measurement, just like the book of OHDSI write like the red box in the picture, it just contains the observation_concept_id,right?

Not sure what you mean. The snippet from the book is basically saying that the model will take in account all the different measurements your cohort’s members had prior to the cohort start date. Then, you can exclude all the measurements (covariates) that occur in a predefined fraction of the target population or less (see Executions settings in ATLAS).

image


(panpan) #5

Thank you very much @George_Argyriou, my question is , for example, if I choose the observation domain, then the included covariates are only the observation_concept_id , the different observations my cohort’s members had prior to the cohort start date, or they include all concept ids, such as observation_ type_concept_id and value_as_concept_id,qualifier_concept_id,unit_concept_id and so on, which are also observed prior to the cohort start date


(George Argyriou) #6

No problem @pandamiao
I see what you mean now. I think it is only the observation_concept_id’s that the model is counting. But you can ask your question again and add a request in the PLP github page.


(panpan) #7

thank you very much!


t