SQL queries and R workflow to create datasets for hourly-adjusted urine output (UO) and KDIGO stages as well as the original research results from the AmsterdamUMCdb. This repository accompanies the article: "Toward the standardization of big datasets of urine output for AKI analysis: a multicenter validation study".
Accurate diagnosis and analysis of oliguric-AKI relies on timely UO charting. The lack of standardization in handling UO data and the various interpretations of KDIGO-UO guidelines limit the ability to make consistent comparisons and draw general conclusions. We aimed to establish a method for standardizing hourly UO using real-life charting data and to examine whether this method can identify oliguric-AKI. We also aimed to validate the method externally.
The model described, based on simple charting, can be used across the board for oliguric-AKI research. It may serve to analyze publicly available DBs and data sourced from standard EHRs as well as custom-made data in Excel tables.
This repository is addressing the validation cohort.
For the derivation cohort see: https://github.com/arielhasidim/mimic-uo-and-aki
This repository has two main objectives:
-
Enable the creation of hourly-adjusted UO and AKI events tables - For further instructions visit:
create_data/...
. -
Enable the reproduction of the associtated article - For further instructions visit:
reproduce_article/...
.
- The AmsterdamUMC database is required in order to run this code and is not provided with this repository.
- To access the AmsterdamUMCdb, you will need to request access according to the guidance on the official website.
- See also the official repository.
- The queries and workflow included here depend on Google's Bigquery cloud service. Researchers who want to use the suggested workflow need to download the AmsterdamUMC dataset tables in full and upload them manually to a Bigquery dataset named "original" within their personal project folder. See the workflow in the screenshot below for reference.
- You will need a Google Cloud Platform (GCS) billing account to run the queries.
- The SQL queries are written in GoogleSQL dialect (formally known as "Standard-SQL" dialect) and is probably compatible with other common dialects.
- The code was tested on AmsterdamUMCdb v1.0.2.
After creating all the tables and reproducing the associated study, you should end up with a result page in HTML format: https://arielhasidim.github.io/aumc-uo-and-aki.
If you use this repository, please cite:
Hasidim, A.A., Klein, M.A., Ben Shitrit, I. et al. Toward the standardization of big datasets of urine output for AKI analysis: a multicenter validation study. Sci Rep 15, 20009 (2025). https://doi.org/10.1038/s41598-025-95535-4
@article{Hasidim2025,
author = {Hasidim, Ariel Avraham and Klein, Matthew Adam and Ben Shitrit, Itamar and Einav, Sharon and Ilan, Karny and Fuchs, Lior},
title = {Toward the standardization of big datasets of urine output for AKI analysis: a multicenter validation study},
journal = {Scientific Reports},
year = {2025},
volume = {15},
number = {1},
pages = {20009},
doi = {10.1038/s41598-025-95535-4},
url = {https://doi.org/10.1038/s41598-025-95535-4}
}