Adaptive Drug Delivery to Control Mean Arterial Blood Pressure by Reinforcement Fuzzy Q-Learning

Sodium nitroprusside (SNP) typically operates in an open-loop configuration, despite the fact that it may successfully regulate blood pressure (BP) in many situations. Because of the medication’s quick action, patients must have a caregiver who can modify the SNP infusion according to each patient’s needs. In order to address this problem, this work presents an auto-update sensor that uses a noninteger controller known as noninteger proportional integrative derivative (Ni-PID) in conjunction with a multiagent fuzzy Q-learning (FQL) algorithm to regulate the BP and drug infusion rate in a closed-loop system. This method explores the use of an FQL agent to achieve the best possible design for the dynamic system. The MABP rate is first detected by a sensor and subsequently transmitted to the Ni-PID. The state value parameter, which is derived from the controller’s performance, is then sent to the FQL algorithm. The outcome of this process is the selection probability density function of the policy network, which is often referred to as the actor value. Under this technique, the FQL maximizes an environment-emitted reward signal to dynamically alter the coefficients embedded in the Ni-PID controller. Multiple examinations, encompassing assessments of work conditions, and stability evaluations, in conjunction with diverse indices, have been executed to authenticate the efficacy and viability of the suggested Ni-PID technique reliant on FQL on a dynamic model linked to the control of pharmaceutical infusion rate. Furthermore, the controllers are incorporated into real-time testbed in order to verify the effectiveness in a practical standpoint.