Achieving optimal antibiotic exposure in critically ill pediatric patients is difficult due to (their) dynamic physiology and variability. Conventional weight-based regimens often fail to reach pharmacokinetic/pharmacodynamic (PK/PD) targets for narrow therapeutic index agents such as vancomycin and amikacin. Model-Informed Precision Dosing (MIPD), which integrates Bayesian forecasting with population pharmacokinetics (popPK), offers a potentially valuable yet underexplored approach in pediatric intensive care to better attain and sustain target exposure. This pilot randomized clinical trial evaluated MIPD-guided dosing of vancomycin and amikacin using InsightRX Nova® versus standard of care (SoC) in a tertiary PICU. Patients whose model-recommended doses matched standard regimens were analyzed under SoC. Primary outcomes included prediction accuracy (a priori vs a posteriori) and model fit; secondary outcomes assessed dose optimization, inflammatory response, renal safety, treatment duration, and mortality.
Study Type
INTERVENTIONAL
Allocation
RANDOMIZED
Purpose
TREATMENT
Masking
NONE
Enrollment
41
It is a precision dosing platform that combines population pharmacokinetic/pharmacodynamic (popPK/PD) modeling with artificial intelligence and machine learning to optimize individualized drug therapy. It uses patient-specific demographic, clinical, and laboratory data to generate real-time personalized dosing recommendations based on validated popPK models.
Hacettepe University
Ankara, Turkey (Türkiye)
Predictive Accuracy of Model-Informed Precision Dosing (MIPD) Based on Median Absolute Error (MdAE)
Predictive accuracy of the pharmacokinetic model will be evaluated by calculating the Median Absolute Error (MdAE) between model-predicted and observed antibiotic serum concentrations. MdAE was prespecified as the primary accuracy metric due to its robustness to outliers in small pediatric samples.
Time frame: From first therapeutic drug monitoring (TDM) sample to second TDM sample (typically within 3-5 days of therapy)
Predictive Accuracy of Model-Informed Precision Dosing (MIPD) Based on Mean Absolute Error (MAE)
Mean Absolute Error (MAE) between model-predicted and observed serum antibiotic concentrations will be calculated to assess overall prediction error.
Time frame: From first therapeutic drug monitoring (TDM) sample to second TDM sample (typically within 3-5 days of therapy)
Predictive Accuracy of Model-Informed Precision Dosing (MIPD) Based on Median Error (MdE)
Median Error (MdE) will be calculated to evaluate directional bias between predicted and observed serum antibiotic concentrations.
Time frame: From first therapeutic drug monitoring (TDM) sample to second TDM sample (typically within 3-5 days of therapy)
Change in C-Reactive Protein (CRP) Level
The absolute change in serum C-reactive protein (CRP) level from baseline (within 24 hours of antibiotic initiation) to the time of the second TDM sample will be evaluated as an inflammatory response marker.
Time frame: Baseline to approximately day 3-5 of therapy
Change in Procalcitonin Level
The absolute change in serum procalcitonin level from baseline to the time of the second TDM sample will be assessed.
Time frame: Baseline to approximately day 3-5 of therapy
Change in Serum Creatinine Level
The absolute change in serum creatinine level from baseline to the time of the second TDM sample will be evaluated as a marker of renal safety.
Time frame: Baseline to approximately day 3-5 of therapy
Change in Pharmacokinetic Model-Fit Category
Within-patient change in pharmacokinetic model-fit category (poor, intermediate, good) between the first and second TDM measurements will be assessed to evaluate improvement in model performance after Bayesian updating.
Time frame: From first to second TDM sample (typically within 3-5 days)
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.