I've been running some large-scale analysis using Strategus--finding the incidence rates (using CohortIncidence module of course) of about 2000 drugs and 800 outcomes. I have 4 TARs, and stratified by age, gender, and index year.
After generating my specification files, I was able to successfully run CohortGenerator. It seems to successfully run CohortIncidence as well. However, it seems to be stuck on the part where it needs to download the results as a CSV onto the Strategus results folder.
(this screenshot was not with the full set of 2000 drugs and 800 outcomes--more like 80 drugs and 800 outcomes. With the full set of 2000 drugs and 800 outcomes, the module did run, but it would get stuck at that step of downloading the CSV.)
What has worked eventually was splitting up the number of outcomes/drugs per run and just looping over my drugs and outcomes until everything is run. Technically, for me, this issue is resolved because I found a workaround, but a potential feature request would be potentially how Strategus deals with these larger-scale studies (and potentially beyond just the CohortIncidence module?). I also noticed my computer eventually does run out of memory when downloading these large CSV files (I have 32gb memory, so it's not super large but it is better than a laptop).
I've been running some large-scale analysis using Strategus--finding the incidence rates (using CohortIncidence module of course) of about 2000 drugs and 800 outcomes. I have 4 TARs, and stratified by age, gender, and index year.
After generating my specification files, I was able to successfully run CohortGenerator. It seems to successfully run CohortIncidence as well. However, it seems to be stuck on the part where it needs to download the results as a CSV onto the Strategus results folder.
What has worked eventually was splitting up the number of outcomes/drugs per run and just looping over my drugs and outcomes until everything is run. Technically, for me, this issue is resolved because I found a workaround, but a potential feature request would be potentially how Strategus deals with these larger-scale studies (and potentially beyond just the CohortIncidence module?). I also noticed my computer eventually does run out of memory when downloading these large CSV files (I have 32gb memory, so it's not super large but it is better than a laptop).