Robust model predictive control for optimal continuous drug administration.

Area of Science:

• Pharmacology and Biomedical Engineering
• Control Systems Engineering
• Computational Biology

Background:

• Optimal drug administration requires precise control to maintain therapeutic levels while avoiding toxicity.
• Existing control methods may not adequately handle system constraints in biological systems.
• Physiologically-based pharmacokinetic (PBPK) models offer detailed dynamic representations of drug behavior.

Purpose of the Study:

• To apply Model Predictive Control (MPC) for optimal drug administration.
• To ensure drug concentrations satisfy minimum toxic concentration (MTC) constraints.
• To develop a robust control system using PBPK models and state estimation.

Main Methods:

• Formulated a whole-body PBPK model as a discrete-time state-space model.
• Designed an artificial observer for real-time estimation of unmeasured drug concentrations.
• Implemented an MPC controller to manage drug dosing based on plasma measurements and estimations.

Main Results:

• The integrated observer and MPC system effectively controlled drug concentrations to target levels in organs.
• System performance was validated with a 7-compartment PBPK model, including constraints on 5 tissues.
• The methodology demonstrated robustness against modeling errors, disturbances, noise, and inaccurate measurements.

Conclusions:

• MPC provides a powerful framework for optimizing drug administration with explicit constraint handling.
• The proposed approach ensures drug safety by adhering to minimum toxic concentration limits.
• This PBPK-model-based MPC strategy is efficient and reliable for drug dosing control applications, even in uncertain environments.