(740a) Model-Based Closed-Loop Controller Design of Multi-Location Vagal Nerve Stimulation for Regulating Cardiovascular System

Cardiovascular disease (CVD) is a major cause of morbidity and mortality, which is related to dysfunction of the autonomic nervous system, with increased sympathetic activity and decrease in vagal tone. In recent years, vagal nerve stimulation has been shown to have beneficial effects for treating CVD in both clinical and preclinical studies. In clinical studies, vagal nerve stimulation (VNS) is delivered in an open-loop approach, determining stimulation parameters based on the patient’s perception. However, the physiological response to a given configuration shows inter and intra-patient variability so that determining the stimulation parameters remains an open challenge.

A number of preclinical studies have proposed closed-loop methods to determine stimulation configurations based on the feedback of heart rate. However, other significant physiological parameters for evaluating cardiovascular performance, such as blood pressure, stroke volume, and respiratory rate are also critical but never controlled. In addition, previous studies have shown the complex structure of the vagal nerve and various responses by stimulating different positions of the vagal nerve, but stimulation position is never considered in the design of the controller. Here we develop a closed-loop framework for multi-location VNS system, which controls heart rate and mean arterial pressure at the same time by independently manipulating stimulation configurations in three different locations.

The proposed framework couples a model predictive control (MPC) algorithm and a real-time simulation of a physiological model representing the cardiovascular response to multi-location VNS. The physiological model incorporates circuit flow mechanics, which represents the interaction of the pump with the left ventricle of the heart and afferent and efferent dynamics of the baroreflex control system. The resulting model is in the form of nonlinear coupled ordinary differential equations which have switching conditions for four consecutive intervals representing the contraction, ejection, relaxation and filling phase of the left ventricle. Both linear and nonlinear MPCs are tested with the proposed physiological model. Our results show the feasibility and usefulness of the proposed closed-loop design for regulation of heart rate and mean arterial pressure. To the best of our knowledge, this is the first systematic attempt to develop a multivariable controller formulation for the VNS problem which has largely been restricted in literature to either open-loop or single-input-single-output closed loop control. Our subsequent work will investigate the real-time implementation of this proposed controller design on an animal model.