On Physiological Closed-Loop Control for Lung-Protective Mechanical Ventilation

Mechanical ventilation remains a cornerstone in the treatment of patients with respiratory failure. However, careful, constant, and lung-protective adjustment of ventilator settings is required to prevent ventilator-induced lung injury. This work presents contributions to the field of physiological closed-loop control, which has the potential to achieve continuous lung-protective mechanical ventilation and reduce clinician workload. First, the large signal behaviour of the respiratory system is studied, including the non-linear hysteretic phenomenon that significantly affects the operating points of tidal ventilation. A mathematical Preisach hysteresis model of the respiratory system is derived, accurately replicating clinically measured pressure-volume curves and serving as a foundation for the automation of lung recruitment manoeuvres. Next, the model-based design of robust closed-loop controllers for precise control of systemic oxygenation is described. Extensive simulations and results from animal experiments demonstrate the controllers' accuracy, reliability, and safety despite significant uncertainties in the process. Finally, a complete physiological closed-loop system capable of automatically adjusting all ventilator settings is presented. Results from animal trials demonstrate that the system can maintain subjects within defined targets while adhering to lung-protective guidelines and requiring minimal clinician interaction.