Determining steady-state trough range in vancomycin drug dosing using machine learning
Authors' conclusion
We developed models to assess the risk of sub- and supra-therapeutic vancomycin trough levels to improve the accuracy of drug dosing in critically ill patients.
Study context
- Population
- Adult intensive care units (ICU) patients (≥ 18 years) with at least one vancomycin concentration measurement, Mayo Clinic, Rochester, MN, January 2007 to December 2017
- Participants
- 5,337 vancomycin courses
- Exposure
- XGBoost vs other machine learning classification models
- Outcome
- Prediction of sub-, supra-therapeutic vancomycin trough levels (classification performance)
Methodological detailsmachine learning prediction model development, single-centre, comparison of classification models
- Machine learning prediction model development
- Single-centre
- Comparison of classification models
Main finding
As reported in the abstract
XGBoost model performance
| Group | AUC-ROC | Sensitivity | Specificity |
|---|---|---|---|
| Sub-therapeutic | 0.85 | 94% | 53% |
| Supra-therapeutic | 0.83 | 94% | 47% |
AUC-ROC
Sub-therapeutic0.85
Supra-therapeutic0.83
Sensitivity
Sub-therapeutic94%
Supra-therapeutic94%
Specificity
Sub-therapeutic53%
Supra-therapeutic47%
Observational data: associations, not effects.
Further findings
Most important model variables
- Kinetic eGFR and other creatinine-based measurements, vancomycin regimen (dose and interval), comorbidities, BMI, age, sex and BP among the most important variables
Original publication
Determining steady-state trough range in vancomycin drug dosing using machine learning. J Crit Care. Published 18 March 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →