Using hospital labs for ICU prediction tasks?

[rmovva]

Using hospital labs for ICU prediction tasks?

There are duplicate values for same measurement at same time points in lab and chart events. As per MIMIC documentation if duplicates we should pick value from labevents. We will explore data in both lab and chart events and implement the strategy suggested in MIMIC. We will also try to document what values are mostly duplicated in both. Thanks, Mehak

On Tue, Jan 31, 2023 at 13:50 'Raj Movva' via healthylaife < ***@***.***> wrote: Thanks Mehak! I didn't know that there are many duplicates between chartevents and labevents, thanks for pointing that out. Are specific measurements duplicated (e.g. a patient's blood results are copied into both chartevents and labevents), or is it that for a particular measurement (e.g. Glucose) might have some values in chartevents and some (other) values in labevents at different timepoints for a given patient? Anyway, I appreciate your efforts. If chartevents has much of the data from labevents, perhaps I can start with that while waiting for your new implementation (no rush). Thank you. — Reply to this email directly, view it on GitHub <#33 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AUP6H74HITLUTRUZV7ZECEDWVFNHLANCNFSM6AAAAAAUMVURSU> . You are receiving this because you commented.Message ID: ***@***.***>

-- Thanks, Mehak Department of Computer and Information Sciences University of Delaware

Dear Raj, Thank you very much for sharing your suggestions with us! We really appreciate that. Hi Mehak, Can you (or Sakhee) please look into integrating Raj's suggestion into the pipeline? Rahmat

…

------------------------------ Rahmatollah Beheshti, PhD | sites.udel.edu/rbi Assistant Professor, University of Delaware Computer & Info. Sciences Department, Epidemiology Program (Joint), Data Science Institute Research Faculty, Nemours Children's Health

On Tue, Jan 31, 2023 at 11:55 AM 'Raj Movva' via healthylaife < ***@***.***> wrote: Hi, Thank you for this useful pipeline. One suggestion: as you probably know, many papers use hospital lab measurements to predict outcomes of ICU stays. In particular e.g. this paper <https://arxiv.org/pdf/2102.06761.pdf> uses many items in the "hosp/labevents.csv.gz" file as features for the ICU stay (linking via the hospital admission ID, "hadm_id", which is a column in the ICU stay matrix). However, I noticed that your pipeline does not natively allow the user to include lab events if they select the ICU flag. It shouldn't be too hard to support that with the code you've already written, so just wanted to flag this point. Let me know if I'm misunderstanding something and this is already an option. Thanks! — Reply to this email directly, view it on GitHub <#33>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AUP6H766LDRJA4CGQCSPKKLWVE7ZBANCNFSM6AAAAAAUMVURSU> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

Dear Raj, Thank you for sharing your suggestions with us. Currently our pipeline allows user to include chartevents in ICU admissions which consists of lab and vital values. There are many duplicates in chartevents and labevents for lab values. We will work to remove duplicates as recommended by MIMIC documentation and refer the paper shared by you in detail to include labevents for ICU admissions. Thanks again for your valuable suggestions. It will help improve our current work. Thanks, Mehak

On Tue, Jan 31, 2023 at 12:10 Rahmat Beheshti ***@***.***> wrote: Dear Raj, Thank you very much for sharing your suggestions with us! We really appreciate that. Hi Mehak, Can you (or Sakhee) please look into integrating Raj's suggestion into the pipeline? Rahmat ------------------------------ Rahmatollah Beheshti, PhD | sites.udel.edu/rbi Assistant Professor, University of Delaware Computer & Info. Sciences Department, Epidemiology Program (Joint), Data Science Institute Research Faculty, Nemours Children's Health On Tue, Jan 31, 2023 at 11:55 AM 'Raj Movva' via healthylaife < ***@***.***> wrote: > Hi, > > Thank you for this useful pipeline. > One suggestion: as you probably know, many papers use hospital lab > measurements to predict outcomes of ICU stays. In particular e.g. this > paper <https://arxiv.org/pdf/2102.06761.pdf> uses many items in the > "hosp/labevents.csv.gz" file as features for the ICU stay (linking via the > hospital admission ID, "hadm_id", which is a column in the ICU stay > matrix). However, I noticed that your pipeline does not natively allow the > user to include lab events if they select the ICU flag. > > It shouldn't be too hard to support that with the code you've already > written, so just wanted to flag this point. Let me know if I'm > misunderstanding something and this is already an option. Thanks! > > — > Reply to this email directly, view it on GitHub > <#33>, or > unsubscribe > <https://github.com/notifications/unsubscribe-auth/AUP6H766LDRJA4CGQCSPKKLWVE7ZBANCNFSM6AAAAAAUMVURSU> > . > You are receiving this because you are subscribed to this thread.Message > ID: ***@***.***> > --

Thanks, Mehak Department of Computer and Information Sciences University of Delaware

[rmovva]

Thanks Mehak! I didn't know that there are many duplicates between chartevents and labevents, thanks for pointing that out. Are specific measurements duplicated (e.g. a patient's blood results are copied into both chartevents and labevents), or is it that for a particular measurement (e.g. Glucose) might have some values in chartevents and some (other) values in labevents at different timepoints for a given patient?

Anyway, I appreciate your efforts. If chartevents has much of the data from labevents, perhaps I can start with that while waiting for your new implementation (no rush). Thank you.