Model Predictive Valve Control to Assist Lung Pressure Profile Tracking

Abstract

In U.K. 60 000 people have a brain tumor and typically are unaware of its presence until symptoms occur. Currently, there is no mass screening available due to limitations in diagnostic techniques. Measurement of intracranial pressure (ICP) [via tympanic membrane displacement (TMD)] is a potential low-cost, accessible solution; however, pressure fluctuations degrade its accuracy. This article solves the problem by assisting participants to precisely track airway pressure profiles. This stabilizes intrathoracic pressure, significantly reducing the fluctuations and enabling accurate diagnosis of ICP. This article develops and evaluates the first model of lung pressure tracking to embed volitional control action. A clinically feasible identification approach is then derived, together with a novel model predictive control (MPC) framework, embedding a valve control subsystem. Results with ten participants confirm that tracking is improved by an average of 22%.