Considering the date shift procedure adopted in MIMIC-III (add a random offset for each patient), is there any way to identify ordered admissions or ordered patients records? My intention is to split the dataset into two even halves, to be able to compare them, however, it must be ordered for this purpose.


See https://github.com/MIT-LCP/mimic-code/issues/442

If you simply want two large groups and knowledge that one group was admitted before the other, then it's feasible with the dataset as-is. You can split the data using the dbsource column in icustays. One group can be dbsource = 'carevue', and the second group can be everyone else. The CareVue group were admitted first.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.