ICU sedation, delirium & recovery
Development and validation of the creatinine clearance predictor machine learning models in critically ill adults
Authors' conclusion
Prediction models based on routinely collected clinical data in the ICU were able to accurately predict next-day CrCl.
Study context
- Population
- Critically ill adults; development in EPaNIC multicentre RCT database, external validation at University Hospitals Leuven (M@tric database)2,825 development, 9576 validation
- Comparison
- Machine-learning models predicting next-day creatinine clearance (CrCl) vs reference assuming the same CrCl as the day of prediction
- Outcome
- Prediction error vs actual CrCl one day ahead (MAE and RMSE)
- Follow-up
- Next day
Methodological detailsgradient boosting machine learning, multicentre development cohort
- Gradient boosting machine learning
- Multicentre development cohort
Performance
As reported in the abstract
Prediction error, external validation cohort
| Test or model | Mean absolute error (MAE) | Root-mean-square error (RMSE) |
|---|---|---|
| Reference | 20.6 (95% CI 20.3-20.9) | 40.1 (95% CI 37.9-42.3) |
| Core + BGA + Monitoring | 18.1 (95% CI 17.9-18.3) | 28.9 (95% CI 28-29.7) |
Mean absolute error (MAE)
Reference20.6 (95% CI 20.3-20.9)
Core + BGA + Monitoring18.1 (95% CI 17.9-18.3)
Root-mean-square error (RMSE)
Reference40.1 (95% CI 37.9-42.3)
Core + BGA + Monitoring28.9 (95% CI 28-29.7)
Original publication
Development and validation of the creatinine clearance predictor machine learning models in critically ill adults. Crit Care. Published 6 July 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →