Deep learning approaches for time series prediction of renal recovery in medical critically Ill patients with acute kidney injury: Long short-term memory (LSTM), gated recurrent unit (GRU), and transformer models
Authors' conclusion
Transformer with LOCF-TG-M accurately predicted 7-day renal recovery in critically ill patients with AKI and identified clinically meaningful predictors to support individualized management.
Study context
- Population
- Adult medical intensive care unit (ICU) patients with acute kidney injury (AKI) admitted between December 2022 and February 2025108 temporal validation
- Participants
- 438 patients
- Exposure
- Transformer vs GRU vs LSTM
- Outcome
- Prediction of renal recovery (model performance)
- Follow-up
- 7 days
Methodological detailsprognostic model development, temporal validation, deep learning
- Prognostic model development
- Temporal validation
- Deep learning
Main finding
As reported in the abstract
Prediction of 7-day renal recovery
| Group | AUROC | F1 score | Accuracy |
|---|---|---|---|
| Transformer | 0.97 | 0.89 | 0.91 |
| GRU | 0.96 | — not stated in the abstract | — not stated in the abstract |
AUROC
Transformer0.97
GRU0.96
F1 score
Transformer0.89
GRU— not stated in the abstract
Accuracy
Transformer0.91
GRU— not stated in the abstract
— not stated in the abstract
Observational data: associations, not effects.
Further findings
Renal recovery incidence · 7 days
- Patients with renal recovery
- 162 (37%)
Key predictors (SHAP analysis)
- Lower AKI stage, sepsis-associated AKI, cardiorenal syndrome, higher urine output, lower shock index, lower serum potassium and blood urea nitrogen, lower comorbidity, higher SpO₂/FiO₂ ratio, lower mean airway pressure, higher dynamic compliance
Original publication
Deep learning approaches for time series prediction of renal recovery in medical critically Ill patients with acute kidney injury: LSTM, GRU, and transformer models. Crit Care. Published 7 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →