Real-time tracking of breathing parameters in mechanically ventilated dogs

A recursive identification algorithm with a constant forgetting factor is applied to online identification of the mechanical properties of the respiratory system. The well-known two-parameter model, accounting for pulmonary compliance and resistance, is adopted to characterize breathing mechanics. The algorithm is first tested by using flow and pressure data measured on mechanically ventilated dogs in time-invariant physiopathological conditions. Subsequently, its tracking ability is evaluated on the basis of numerically simulated data in a realistic and well-defined time-varying situation. The results obtained prove that a good tradeoff between tracking ability and noise sensitivity can be reached in the presence of realistic noise levels. The algorithm's computational efficiency allows real-time tracking of the slow parameter changes typical of respiratory pathologies.