Bayesian Dosing Simulator (BDS): A Pharmacokinetic Modeling Tool for Optimized Antibiotic Therapy

Area of Science:

• Pharmacokinetics and Pharmacodynamics
• Computational Biology
• Infectious Disease Therapeutics

Background:

• Meropenem efficacy for hospital-acquired pneumonia relies on maintaining drug concentrations above the minimum inhibitory concentration (MIC).
• Standard therapeutic drug monitoring (TDM) may be insufficient for target attainment in critically ill patients, especially those on continuous renal replacement therapy (CRRT).
• Model-informed precision dosing (MIPD) offers potential for individualized meropenem dosing but faces implementation challenges.

Purpose of the Study:

• To develop and validate a Shiny application for Bayesian meropenem dose optimization.
• To implement a previously published nonparametric model within a parametric Bayesian framework using maximum a posteriori (MAP) updating.
• To assess the predictive performance of the developed application using pharmacokinetic data from critically ill patients.

Main Methods:

• Translated a nonparametric meropenem model into a parametric Bayesian framework with MAP updating via a Shiny interface.
• Validated the model using plasma pharmacokinetic data from critically ill patients, comparing predictions with Pmetrics and Monte Carlo simulations.
• Assessed predictive performance using relative median prediction error (rMPE) and relative median absolute prediction error (rMAPE), evaluating sparse sampling scenarios.

Main Results:

• The Bayesian Dosing Simulator (BDS) demonstrated robust performance across sparse sampling scenarios (1-3 observations), with prediction errors within prespecified limits (rMPE ±20%, rMAPE ≤30%).
• F20 (probability of achieving target exposure) ranged from 80% to 94%, and F30 ranged from 87% to 94%.
• The developed application showed predictive accuracy comparable to reference models, outperforming simulation-based population predictions, particularly in CRRT patients.

Conclusions:

• A validated nonparametric meropenem model was successfully implemented in an open-source Bayesian framework, enabling accurate, individualized dosing.
• The application demonstrated robust performance under sparse sampling conditions, supporting its feasibility for clinical use.
• Further prospective studies are required to evaluate the clinical safety and effectiveness of this MIPD approach for meropenem therapy.