Optimal adaptive control of drug dosing using integral reinforcement learning

Area of Science:

• Anesthesiology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Maintaining optimal sedation levels during procedures is critical in intensive care units (ICUs).
• Traditional methods for sedative drug administration may lack precision and adaptability to individual patient responses.
• Propofol is a commonly used anesthetic agent requiring careful dosage management.

Purpose of the Study:

• To develop and evaluate a reinforcement learning (RL)-based optimal adaptive control strategy for continuous sedative drug infusion.
• To ensure precise maintenance of target sedation levels in patients.
• To adapt drug dosing dynamically based on individual patient pharmacology.

Main Methods:

• An online integral reinforcement learning (IRL) algorithm was designed for optimal control.
• The controller iteratively updates drug dosing strategies.
• The system was tested using 10 simulated patients receiving propofol infusion.

Main Results:

• The proposed IRL-based controller demonstrated efficacy in maintaining target sedation levels.
• The adaptive control approach successfully adjusted propofol dosing based on simulated patient pharmacology.
• Numerical simulations confirmed the controller's ability to converge to optimal solutions.

Conclusions:

• Reinforcement learning offers a promising approach for optimal adaptive control of sedative drug infusion.
• The IRL algorithm provides a robust method for personalized drug dosing in critical care settings.
• This technology has the potential to enhance patient safety and treatment outcomes in ICUs.