Need help generating a dataset for a specific time period.

[allen-ajith]

Hello, I’m trying to export patient data with exactly 4 years of encounter history. I’ve tried setting exporter.years_of_history = 4 in the properties file and also used the command line argument --exporter.years_of_history=4, but I still see some encounters or observations outside this range.
Is there a way to strictly limit the exported data to only encounters and observations from the last 4 years?

I'm using the command

java -jar synthea-with-dependencies.jar -p 1000 "New York" "New York"
--exporter.csv.export=true
--exporter.fhir.export=false
--exporter.years_of_history=3

[jawalonoski]

No. The years of history is not strict, it will include clinically active concepts outside the time window.

So for example, if a patient has active diabetes and hypertension from 5 years ago, they will be included, as well as current medications they are taking for them.

If you need it to be strict, you have to post-process the data yourself.

[robinsonha]

Will seeding patients we create in one run allow us to come back and generate a new later time period for the same patients continuing their record or does that just create the same patients for different later dates? To add to an existing patient record without disrupting their health trajectory is this possible? Or should I postprocess a lifetime's worth of data to reflect monthly requirements?

[jawalonoski]

Will seeding patients we create in one run allow us to come back and generate a new later time period for the same patients continuing their record or does that just create the same patients for different later dates?

If you use the same version of the software, with the same modules, seeds, etc... yes, it will generate the same patient with all the old data and new data.

To add to an existing patient record without disrupting their health trajectory is this possible?

Except for the above, the only other way to do this is to try this experimental feature: https://github.com/synthetichealth/synthea/wiki/Evolving-a-Population

Or should I postprocess a lifetime's worth of data to reflect monthly requirements?

Many organizations do this. Run data into the future by modifying the end date ([-e endDate as YYYYMMDD]) and performing periodic ETL or other processing to pull in the "new" or "refreshed" data. You can make this more complicated by having several different batches going so that new patients/beneficiaries will appear in the data as well.