Robust closed-loop control of the inspired fraction of oxygen for the online assessment of recruitment maneuvers.

Recruitment maneuvers are commonly used in patients suffering from acute respiratory failure. A continuous measurement of PaO(2) would help to assess the proper execution of such maneuvers. Unfortunately, there are only static offline measurement devices available. However, if the oxygen saturation would be held close to a fixed set-point of 90-92% by automatically adjusting the FIO(2) parameter of the mechanical ventilator, the latter would be inversely related to PaO(2). In this work, a new robust PID control system is presented, which accounts for model uncertainty in a modified Smith predictor approach. The controller was tested using a computer model of respiratory failure. The model parameters were identified in an animal experiment with pigs.