Koopman-based Impulsive Model Predictive Control of BCG Immunotherapy.

Area of Science:

• Computational biology
• Immunotherapy
• Pharmacodynamics

Background:

• Bacillus of Calmette and Guerin (BCG) is a key immunotherapy for non-muscle-invasive bladder cancer.
• Designing optimal BCG dosing is challenging due to nonlinear pharmacodynamics and model constraints.

Purpose of the Study:

• To develop a computational method for designing optimal BCG drug dosing schemes.
• To address limitations in current BCG treatment protocols for bladder cancer.

Main Methods:

• Utilized Koopman theory to linearize nonlinear pharmacodynamic models.
• Employed model predictive control for impulsive drug dosing.
• Incorporated constrained optimization for drug concentration limits.

Main Results:

• The Koopman-based linear model accurately replicated the original nonlinear system dynamics.
• Designed drug doses adhered to specified constraints.
• Effectively suppressed cancerous cell proliferation by guiding tumor cell populations.

Conclusions:

• The developed computational method offers an optimal, impulsive, and linear approach to drug dosing.
• This strategy shows promise for generalizing model-based drug delivery in various treatments.
• The method enhances BCG immunotherapy efficacy for bladder cancer while respecting safety constraints.