In silico evaluation of gas transfer estimation during extracorporeal membrane oxygenation

The application of extracorporeal membrane oxygenation can prevent hypoxia and the inherent danger of death, if despite artificial ventilation maintaining physiological gas exchange is barely possible. In doing so, an oxygenator enables supplemental blood oxygenation and decarboxylation. The automation of this extracorporeal therapy is in the focus of current research. However, only complex and slow blood gas analyzers are available as measurement systems so far. This paper presents the in silico evaluation of an extended Kalman filter, which estimates the gas transfer across the membrane without using a blood gas analyzer. The filter is based on an evaluated grey box model of the extracorporeal circulation, which has been extended with additional measurements in the gas phase. This application is enabled by using standard anesthetic gas monitors at the in- and outlet of the gas phase of the oxygenator, which measure the oxygen and carbon dioxide fractions. These systems are accurate, easily applicable with inexpensive and few standard disposables and mainly limited by a time delay of 5 s. The results suggest that the estimation is possible with sufficient performance concerning the clinical application and this can be used in future automation applications.