I found that the measurements were taken at different times, and I did not know the date of the initial test.In general, measurements that are too far apart from the initial test date are not useful, so how do I know the date of the initial test?

Created by GAO GAO GaoGao199694
@trberg I'll focus on the question 1. I don't know if I misunderstood the question.My understanding is that if a person is diagnosed on May 1, 2020 (although that date has not been published), then we should be looking at data from a month before May 1, and those older records may not have much impact on the diagnosis.For example, in table device_exposure, there is a record of a person using a ventilator in 2010, but this record may not be very helpful for COVID-19 detection a decade later.
@GaoGao199694, To clarify, are you looking at this for question 1 or question 2? Question 1 doesn't require a test date since there isn't a time aspect and we don't include 706163 measurements in that data. Question 2 does require the test date and we include 706163 measurements in those data. Thanks, Tim
@trberg However, In the 'measuremren.csv' table of the training set I find no record of measurement_concept_id = 706163.I really need to know when the patient was tested for SARS-COV-2 so that I can know which data is valid.

Hi @GaoGao199694 , The initial test date will be the first measurement record of a covid test (measurement_concept_id = 706163) in the measurement table for each patient. The synthetic data is not going to be reflective of the concept distribution of the real data, either longitudinally or across the patient population. So you can't rely too much on the synthetic data to get an idea of how much data is available after the initial test date. Let me know if you'd like more info, Thank you, Tim

How do I know the date of the initial test? page is loading…