A Robust Model Predictive Control Approach to Intelligent Respiratory Support

Respiratory support is one of the main components of almost any intensive care therapy. The trade-off between providing adequate oxygenation and removing carbon dioxide on the one hand while at the same time preventing or reducing ventilator-induced lung injuries is a key challenge for any respiratory therapist. Several decision support systems are available in modern medical ventilators to support clinicians with these challenging decisions, where safety is a core requirement. In this context, the present paper aims to contribute to the field by proposing a robust model predictive controller (MPC) to achieve adequate gas exchange by adjusting the minute volume ventilation within safe physiological limits. A physiological nonlinear two-compartment patient model is used in a robust MPC approach, which guarantees the evolution of the partial pressure of end-tidal carbon dioxide despite an unknown but bounded metabolic production rate and the bilinear dynamics. The latter are considered as additive disturbances and rejected by an additional feedback controller. Simulation results based on a physiological model exemplifies the applicability of the proposed approach.