Personalized protective ventilation of intensive care patients using a digital twin

Purpose and goal:

The goal of this project is to provide intensive care clinicians with tools to ensure that mechanical ventilation does not cause harm to the patient’s lungs and diaphragm by developing a physiological model of the respiratory system that can be tuned to the individual patient, a digital twin. The system will be developed using novel AI-techniques in machine learning using data from a Getinge proprietary solution to measure the electrical activation of the diaphragm.

Expected results and effects:

Mechanical ventilation is provided to more than 20 million patients worldwide every year. To ensure that the therapy is safe and effective there is a need to accurately monitor diaphragm and lung distending pressures in the individual patient. In this project we’ll use a digital twin to provide the means for personalized care. A new hybrid physiological and machine-learning model will be developed, for accurate estimation of clinically relevant pressures, and evaluated against patient data.

Approach and implementation:

The project has two partners, one industrial, Getinge and one academic, KTH. Getinge provides the project with clinical knowledge, patient data, performs pre-processing and sets up the physiological lung model. KTH is responsible for machine-learning model construction, training and validation. KTH will also share knowledge, competence and understanding of machine learning to Getinge.