ETODA: Automatic three-dimensional error tolerance grid generation for dosage adaptation in precision medicine

Area of Science:

• Pharmacokinetics and Pharmacodynamics
• Computational Biology
• Clinical Pharmacology

Background:

• Model-informed precision dosing (MIPD) uses drug measurements for individualized regimens.
• Point-of-care (POC) technologies may introduce measurement uncertainty compared to lab assays.
• Existing frameworks lack tools to assess how measurement inaccuracies impact dosing decisions and outcomes.

Purpose of the Study:

• To develop and validate a computational framework, ETODA (error tolerance of dosage adaptation), for quantifying the robustness of dosage decisions under measurement uncertainty.
• To systematically evaluate how inaccuracies in drug concentration measurements propagate to dosage adaptation and affect therapeutic outcomes at different sampling times.

Main Methods:

• ETODA integrates population pharmacokinetic (popPK) models, patient covariates, Bayesian estimation, and Monte Carlo simulations.
• It constructs 3D error-tolerance grids to map measured vs. true concentrations, sampling time, and therapeutic risk.
• The framework was applied to imatinib and vancomycin using simulated steady-state concentration ranges.

Main Results:

• ETODA revealed drug-specific responses to measurement discrepancies and the impact of sampling time on therapeutic classification.
• For imatinib, peak sampling at 4h maximized therapeutic classification (52.24%), while 24h sampling increased inefficacy alarms (53.47%).
• For vancomycin, AUC/MIC monitoring outperformed trough monitoring in achieving therapeutic targets and reducing alarms.

Conclusions:

• ETODA offers a robustness-evaluation layer for popPK and MIPD workflows, mapping measurement error to dose decisions and risk classifications.
• This framework supports more robust MIPD and guides the development of POC monitoring technologies for improved precision dosing.
• It enhances the safety and efficacy of precision dosing by accounting for measurement uncertainty.