Model-Based Design and Experimental Validation of Control Modules for Neuromodulation Devices.

Abstract

The goal of this paper is to propose a model-based control design framework, adapted to the development of control modules for medical devices. A particular example is presented in which instantaneous heart rate is regulated in real-time, by modulating, in an adaptive manner, the current delivered to the vagus nerve by a neuromodulator. The proposed framework couples a control module, based on a classical PI controller, a mathematical model of the medical device, and a physiological model representing the cardiovascular responses to vagus nerve stimulation (VNS). In order to analyze and evaluate the behavior of the device, different control parameters are tested on a "virtual population," generated with the model, according to the Latin Hypercube sampling method. In particular, sensitivity analyses are applied for the identification of a domain of interest in the space of the control parameters. The obtained control parameter domain has been validated in an experimental evaluation on six sheep. A range of control parameters leading to accurate results was successfully estimated by the proposed model-based design method. Experimental evaluation of the control parameters inside such a domain led to the best compromise between accuracy and time response of the VNS control. The feasibility and usefulness of the proposed model-based design method were shown, leading to a functional, real-time closed-loop control of the VNS for the regulation of heart rate.