Machine learning survival analysis for predicting kidney disease progression in patients with acute kidney injury undergoing continuous kidney replacement therapy: An analysis of the LINKA database
Authors' conclusion
This machine learning survival approach highlights a set of critical, readily measurable predictors of ESKD risk
Study context
- Population
- Acute kidney injury (AKI) patients receiving continuous kidney replacement therapy (CKRT) who survived beyond one week after CKRT (LINKA database)6 development + 2 validation hospitals
- Participants
- 1,444 patients
- Comparison
- CoxBoost vs Elastic-Net Cox vs random survival forest vs Cox proportional hazards models
- Outcome
- Prediction of ESKD, assessed by C-index
Methodological detailsmachine learning survival models, multicentre, external validation, Development: six hospitals; validation cohort: two hospitals; 122 clinical and laboratory variables
- Machine learning survival models
- Multicentre
- External validation
- Development: six hospitals; validation cohort: two hospitals; 122 clinical and laboratory variables
Performance
As reported in the abstract
Prediction of ESKD (C-index)
| Test or model | Concordance index (C-index), internal validation (95% CI 0.756-0.865) | C-index, external validation (95% CI 0.700-0.788) |
|---|---|---|
| CoxBoost | 0.811 | 0.742 |
CoxBoost
Concordance index (C-index), internal validation (95% CI 0.756-0.865)0.811
C-index, external validation (95% CI 0.700-0.788)0.742
Further findings
Key predictors (CoxBoost)
- Variable set reduced to key parameters; most critical: 24-h urine output on day 7 of CKRT, preexisting CKD, day 7 kidney function and systemic laboratory measures
- 23
Simplified scoring system (six binarized variables)
- Effectively stratified patients into low-, intermediate-, and high-risk groups for end-stage kidney disease (ESKD) progression
Original publication
Machine learning survival analysis for predicting kidney disease progression in patients with acute kidney injury undergoing continuous kidney replacement therapy: An analysis of the LINKA database. J Crit Care. Published 24 December 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →