Cooperative Control of ECMO
Fig. 1: System configuration of cooperative control of ECMO and mechanical ventilator.
Contact person
Project description
Whenever artificial ventilation is barely possible to maintain physiological gas exchange, extracorporeal membrane oxygenation (ECMO) is usually applied to prevent hypoxia and the inherent danger of death. The role of ECMO is to provide supplemental gas exchange and to relieve patients from possible excessive stress introduced by a ventilator. The adjustment of ECMO settings must be conformed with general clinical guidelines, but typically in an irregular time interval. Thus, no continuous intra- and interindividual therapeutic optimization is supported.
Project goals
To resolve this issue, the automation is introduced for cooperative control of ECMO and mechanical ventilation in this research work. A traditional closed-loop system relied only on a complex and slow blood gas analyzer for measuring control variables. A new control system is configured as shown in Fig. 2 by adding two standard anesthetic gas monitors in order to measure oxygen (O2) and carbon dioxide (CO2) fractions at the in- and outlet of the gas phase of the oxygenator. Moreover, an extended Kalman filter is applied to estimate O2 and CO2 gas transfer rates across the membrane with no reliance to the complex blood gas analyzer. In addition, we could achieve similar measurement accuracy compared to the standard arterial blood gas analysis with the samples taken before and after the oxygenator simultaneously. Using gas transfer as control variables, safe and reliable ECMO control is realizable.
For implementing the cooperative control, a robust and decentralized control system has been set up to adjust the extracorporeal gas transfer rates according to the demand of the patient. Thus, performance tracking is evaluated based on physiological target control by cooperative adjustments of the ventilator and ECMO control system. Finally, the overall algorithmic performances of this embedded real time system are assessed by porcine models of induced lung injury with surfactant wash-out using repetitive lung lavages.