A virtual patient of extracorporeal membrane oxygenation combining clinical data, numerical simulations, and surrogate modeling

• Applicants: Professor Dr. Christian Karagiannidis; Dr. Michael Neidlin
• Subject Area: Biomedical Systems Technology ; Pneumology, Thoracic Surgery
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 557450644

Extracorporeal membrane oxygenation (ECMO) provides blood oxygenation, decarboxylation, and circulatory support through blood pumps and membrane oxygenators. It is used in the intensive care unit (ICU) for patients with severe respiratory and cardiac failure. Despite its life-saving potential, mortality rates remain unacceptably high, indicating the need for a deeper understanding of ECMO therapy. To improve outcomes, it is essential to evaluate the physiology of the cardiorespiratory system, the hemodynamics within the devices, and the patient-device interaction during ECMO. Current computational tools fail to provide such a comprehensive description and do not effectively utilize the growing amounts of clinical data available in ICUs. To address this challenge, we will create a virtual patient model of extracorporeal membrane oxygenation. This model will combine electronic health records (EHRs), high-fidelity numerical simulations, and data-driven surrogate modeling to enhance our understanding and application of ECMO therapy. Initially, we will expand an existing cardiorespiratory system model during ECMO and implement Bayesian inference for robust parameter estimation with uncertainty quantification. Then, we will use large-scale clinical data from EHRs of more than 700 ECMO patients to perform phenotyping and generate virtual cohorts. High-fidelity computational fluid dynamics will be employed to derive surrogate models of the most common ECMO devices, which will be integrated within the cardiorespiratory model to form the virtual patient. Using this virtual patient model and virtual cohort data, we will conduct a proof-of-concept in-silico clinical trial to evaluate various ECMO circuit configurations and devices. Our methods aim to unravel the complexities of ECMO therapy, paving the way towards digital twin applications in cardiopulmonary intensive care. This framework will support the evaluation of new ECMO treatment strategies and aid in the design and testing of medical devices, ultimately improving patient outcomes.

• DFG Programme: Research Grants