Validate survival variables

[ielis]

Currently we take survival info from CDA.

This is what should be done. The method to calculate the survival time in days depends on the vital status (Alive/Deceased).

• Deceased
  • Subject.days_to_death: Number of days between the date used for index and the date from a person's date of death represented as a calculated number of days.
• Alive - we must calculate the difference between age at last encounter and diagnosis date
  • Diagnosis.age_at_diagnosis: The age in days of the individual at the time of diagnosis.
  • Subject.days_to_birth: Number of days between the date used for index and the date from a person's date of birth represented as a calculated negative number of days.

Alternatively, we can fetch the survival data directly from GDC, when we just fetch two columns:

• vital status (Alive/Deceased)
• survival in days - the number of days after making the diagnosis (index date?)

The GDC API

[justaddcoffee]

@sujaypatil96 how hard would it be to incorporate this into the code that does the call to GDC?

[msierk]

It would seem to be easier to just use the GDC API since they have already calculated the surival time and provided the censoring info. We should get the censored and survival estimate values. We should probably do the calculation from the CDA data and compare, but initially this gives us data we can do a Cox regression with to move forward with the analysis part of the manuscript.

[justaddcoffee]

@sujaypatil96 said he would have a try at pulling survival time from GDC

[sujaypatil96]

I'll spend some time looking into this next week. I'm OOO 4/4 and 4/5.

[justaddcoffee]

Sure, thanks @sujaypatil96

[sujaypatil96]

I have figured out how to pull in survival time (in days) and vital status information from GDC via the /analysis/survival endpoint. Now we just need to stick into the phenopacket on VitalStatus. Perhaps we can do this again some testing on a mini hacking session @ielis @justaddcoffee?

[justaddcoffee]

great progress

@sujaypatil96 @ielis what about a hacking session Fri 2 pm Eastern US time on my zoom?

[sujaypatil96]

I won't be able to make the hacking session on friday because I'll be at an all day training session on thursday, 4/11 and friday 4/12 this week.

But as far as the task goes I think I've completed pulling in survival times and vital status information from GDC and plugged them into the phenopackets as well. It would be nice if you could test it once and let me know.

[sujaypatil96]

Notes from discussion with @justaddcoffee on 4/16.

If person/patient is "Alive" then we need a way of populating survival time in days, and we can do that as follows:

when we last saw the patient (day of last encounter) - when the patient was diagnosed