Multiple Model Predictive Control of the Cardiovascular System Using Vagal Nerve Stimulation

Abstract

Vagal nerve stimulation (VNS) is currently under investigation for the treatment of various cardiovascular diseases including heart failure, arrhythmia, and hypertension. In preclinical and clinical studies, VNS stimulation parameters are heuristically determined in the open loop. However, its therapeutic efficacy remains inconclusive, strongly suggesting the need for a closed-loop approach to optimize patient-specific stimulation parameters. In this paper, we develop a multiple model predictive control (MMPC) algorithm for automated regulation of heart rate (HR) and mean arterial pressure by optimally adjusting the amplitude and frequency of electrical pulses applied to three locations of the vagal nerve. The multiple local models are identified from our previously reported pulsatile rat cardiac model that emulates symptoms of hypertension in rest and exercise states. The computational expense of the proposed method is verified in simulation with rigorous hardware-in-the-loop (HIL) implementation.