Nonlinear Model Predictive Control of Vagal Nerve Stimulation to regulate hemodynamic variables

Abstract

Various pre-clinical investigations indicate that the electrical stimulation of the cervical branch of the vagus that innervates the heart has therapeutic value in the management of various cardiac diseases. In theory, the design of a closed-loop control mechanism that automatically adjusts vagal nerve stimulation (VNS) parameters based on real-time physiological feedback can eliminate intra-patient variability in VNS outcomes and therefore represents a major step towards patient-specific therapy. This study develops a nonlinear model predictive control (NMPC) approach for VNS of a pulsatile, human cardio-baroreflex system. The manipulated variables are the frequency and amplitude of a charge-balanced biphasic current. The effects of these variables on hemodynamic quantities such as heart rate, blood pressure, heart contractility e.t.c. are estimated under the assumption that the desired activation of efferent vagal nerve fibers within the vagosympathetic complex can not be realistically isolated from the off-target activation of afferent fibers. An approximate, cycle-averaged cardiovascular model is derived to eliminate pulsatility and is used for prediction in the controller. The feasibility of this NMPC scheme is explored with a set-point tracking example.