Deep-learning time-series anomaly detection of acute kidney injury from creatinine–eGFR trajectories in the ICU

Yoonjin Kang, Soojeong Yun, Seung Min Song, Ji Eun Kim, Hyo Jin Kim, Eun Jung Cho, Shin Young Ahn, Young Joo Kwon and Min Woo Kang

PLOS Digital Health, 2026, vol. 5, issue 5, 1-19

Abstract: Acute kidney injury (AKI) is common in the intensive care unit (ICU), and fixed creatinine thresholds may miss clinically relevant dynamics. We tested whether a deep-learning anomaly signal from short creatinine–estimated glomerular filtration rate (eGFR) series complements conventional criteria for risk stratification. Seven-step daily creatinine–eGFR instances were built from the Medical Information Mart for Intensive Care (MIMIC-III/IV; development/internal validation) and the eICU Collaborative Research Database (external validation). Time series began within 48 hours before ICU admission and ended at kidney replacement therapy (KRT), death, or ICU discharge. An unsupervised Anomaly Transformer trained on MIMIC produced final-step anomaly scores; the 95th percentile of training scores defined a fixed threshold. Anomaly-detected AKI required a final-step creatinine rise plus a score ≥ threshold. We compared scores across Kidney Disease: Improving Global Outcomes (KDIGO) stages and evaluated 24–96-hour KRT and mortality using area under the receiver operating characteristic curve (AUROC), accuracy, and F1-score. After exclusions, the internal dataset included 81,876 admissions (381,700 time-series instances) and the external dataset 140,237 admissions (494,684 instances). Anomaly scores increased stepwise across KDIGO categories and were higher in windows followed by KRT or death. For KRT prediction at 24, 48, 72, and 96 hours, AUROCs were 0.83, 0.82, 0.81, and 0.80 internally and 0.74 at all horizons externally. For mortality, AUROCs were 0.64–0.66 internally and 0.62–0.64 externally. In threshold-based classification, F1-scores were generally highest with the anomaly rule alone, whereas accuracy was greatest when requiring both anomaly detection and KDIGO stage ≥2. Event-capture analyses showed that anomaly detection identified more near-term KRT and mortality events than KDIGO stage ≥2, with the clearest separation for KRT. A creatinine–eGFR trajectory-based anomaly signal aligned with clinical severity, was associated with near-term outcomes, and appeared to complement KDIGO-based criteria in ICU populations.Author summary: We look after many critically ill patients whose kidneys suddenly become damaged. Doctors usually diagnose this “acute kidney injury” by asking whether a blood test called creatinine crosses fixed cut-off values. But in real ICU care, those rules can miss what we call “hidden acute kidney injury”: clinically consequential kidney injury that may not be flagged by fixed creatinine thresholds at the time of assessment because of baseline uncertainty, dilutional effects, or trajectory heterogeneity. In this study, we asked whether a computer model could uncover these hidden cases by watching how kidney tests change over time, rather than checking a single number. Using two large ICU databases, we fed the model short daily series of creatinine and an estimate of how well the kidneys filter blood. The model learned what typical trajectories look like and then gave each series an “anomaly” score that reflects how unusual it is. We found that higher anomaly scores were linked to a greater chance of starting dialysis or dying within the next few days, even when standard criteria were not met at the same assessment time point. Our results suggest that this trajectory-based signal could complement current definitions and help clinicians recognize hidden acute kidney injury.

Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001411 (text/html)
https://journals.plos.org/digitalhealth/article/fi ... 01411&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pdig00:0001411

DOI: 10.1371/journal.pdig.0001411

Access Statistics for this article

More articles in PLOS Digital Health from Public Library of Science
Bibliographic data for series maintained by digitalhealth ().