Recruitment–Derecruitment in Patient-Specific Respiratory Digital Twins: A Prospective Clinical Validation Study
Please login to view abstract download link
Severe pulmonary disorders such as acute respiratory distress syndrome (ARDS) exhibit pronounced spatial heterogeneity and highly patient-specific mechanics, challenging the design of lung-protective strategies for mechanical ventilation. Physics-based respiratory digital twins promise to replicate individual lung mechanics and serve as a decision support at the bedside. A key requirement is the ability to capture nonlinear and time-dependent phenomena such as alveolar recruitment and derecruitment, which are central to ventilator-induced lung injury [1]. We present results from a prospective clinical pilot study in ten mechanically ventilated ARDS patients, evaluating a patient-specific, regionally resolved, reduced-order lung model including recruitment–derecruitment dynamics [2]. This extends the range of a prior regional modeling and EIT-based validation study [3], as it includes recruitment-derecruitment effects in a novel model. For each patient, the model was calibrated using routine clinical data from ventilator measurements and end-expiratory CT scans. The individualized digital twins were used to predict global respiratory mechanics and regional ventilation across systematically varied ventilator settings, including changes in positive end-expiratory pressure. Model predictions were compared against clinical reference measurements such as tidal volumes, transpulmonary pressures, and regional ventilation distributions assessed by electrical impedance tomography (EIT). To improve the spatial fidelity of the EIT-based regional validation, we employed an anatomically improved EIT reconstruction approach [4]. The model shows strong agreement with clinical data at both global and regional levels, accurately reproducing nonlinear pressure–volume behavior and patient-specific ventilation redistribution with changing ventilator settings. These results highlight the model’s ability to simulate recruitment–derecruitment processes and their impact on regional ventilation. The current level of regional validation is limited by the spatial resolution of EIT. Recent advances in four-dimensional computed tomography (4D-CT) synchronized with respiratory motion enable highly resolved assessment of lung deformation and ventilation [5]. Beyond presenting the clinical validation results, we outline ongoing work toward high-resolution model validation using 4D-CT to further assess locoregional lung mechanics.
