Parameter estimation of recruitment models in mechanical ventilation

Abstract

Intratidal recruitment/derecruitment (R/D) of alveolar space is an important phenomenon which may cause considerable lung injury. Simple models have been proposed to describe pressure [1] or time [2] dependent recruitment. Currently these models are not yet evaluated in praxis i.e. direct comparison of patient data and model output was not performed. But an automatic fitting of measured respiratory data may provide insight into physiological processes and valuable information about the pathological state of the lung at the bedside. Respiratory data obtained from acute lung injury (ALI/ARDS) patients (N=23) and animals (N=8) under mechanical ventilation was fitted on the basis of a combined recruitment model including both approaches described above [1,2]. With a reduced number of parameters the fitting process itself is fast and robust. Reasonable parameter settings were found e.g. describing the influence of the "superimposed" pressure generated by oedema or values for opening and closing pressures. It seems that fitting of recruitment models may provide valuable information about the internal state of a critically ill lung. Verification of internal parameters is currently accomplished in animal experiments. In alveolar microscopy opening pressures can be directly accessed and compared to the fitted internal parameters of the simulation. Parameter estimation of internal i.e. hidden, non-observable parameters of recruitment models may prove to be useful to guide ventilation therapy to avoid intratidal R/D processes. It is just depending on respiratory data obtainable at the bedside without noteworthy effort compared to imaging techniques. Temporally delayed R/D may account for many phenomena solely denoted to viscoelastic properties of the tissue or inhomogeneity of the lung. Future research is devoted to the issue if R/D is partly responsible for these observations. © 2009 Springer Berlin Heidelberg.