Patient-specific prediction of regional lung mechanics in ARDS patients with physics-based models: a validation study

TL;DR

The study validates physics-based, patient-specific computational lung models for ARDS by comparing predictions of regional ventilation to bedside Electrical Impedance Tomography across seven patients. Using a CT-derived anatomy plus ventilator waveforms, the reduced-dimensional model predicts time-resolved regional ventilation and tissue strain, mapped to virtual EIT via electrodynamics for direct comparison. Results show very high agreement in anteroposterior ventilation profiles (Pearson around 0.96) and strong overall concordance across the transverse plane and tidal range (average correlations around 0.81–0.82 with RMSE near 0.14–0.15), supporting the feasibility of model-informed, patient-specific ventilation decisions. The work lays a foundation for mechanistically grounded biomarkers and in silico testing to optimize lung-protective settings in ARDS, with future directions including 4D-CT validation and extension to complex phenomena such as recruitment and derecruitment.

Abstract

The choice of lung protective ventilation settings for mechanical ventilation has a considerable impact on patient outcome, yet identifying optimal ventilatory settings for individual patients remains highly challenging due to the inherent inter- and intra-patient pathophysiological variability. In this validation study, we demonstrate that physics-based computational lung models tailored to individual patients can resolve this variability, allowing us to predict the otherwise unknown local state of the pathologically affected lung during mechanical ventilation. For seven ARDS patients undergoing invasive mechanical ventilation, physics-based, patient-specific lung models were created using chest CT scans and ventilatory data. By numerically resolving the interaction of the pathological lung with the airway pressure and flow imparted by the ventilator, we predict the time-dependent and heterogeneous local state of the lung for each patient and compare it against the regional ventilation obtained from bedside monitoring using Electrical Impedance Tomography. Excellent agreement between numerical simulations and experimental data was obtained, with the model-predicted anteroposterior ventilation profile achieving a Pearson correlation of 96% with the clinical reference data. Even when considering the regional ventilation within the entire transverse chest cross-section and across the entire dynamic ventilation range, an average correlation of more than 81% and an average root mean square error of less than 15% were achieved. The results of this first systematic validation study demonstrate the ability of computational models to provide clinically relevant information and thereby open the door for a truly patient-specific choice of ventilator settings on the basis of both individual anatomy and pathophysiology.

Figures (11)

• Figure 1: Conceptual overview - validation of computational lung model against clinical reference data. Reference: Using an electrode belt positioned around the thorax of the patient, we can observe the regional ventilation during mechanical ventilation (EIT). Simulation: On the basis of a computed tomography (CT) scan we construct a patient-specific model of the lung and conduct a biomechanical and electrodynamic simulation, mimicking both the mechanical ventilation process and the propagation of electrical currents across the patient's torso. The simulated voltages obtained from the computational lung model yield a virtual EIT, which we compare against the bedside EIT obtained from clinical, measured voltage data for a series of timesteps $t$.
• Figure 2: Exemplary transverse CT slices at the level of the EIT belt between $4th$ or $5th$ intercostal space defining the EIT reconstruction plane acquired in three of the studied patients. Markups indicate (yellow) ground-glass opacities, (magenta) dorsobasal atelectasis (with accompanying pleural effusion) and (cyan) pulmonary infiltrates.
• Figure 3: With a high-resolution CT scan as input, the generation of patient-specific models is a streamlined process starting with the segmentation of the voxel data into airways, lungs, and lobes. Beyond the resolution limitations of the CT, a morphology-based tree growth algorithm creates the remaining generations of conducting airways in a space-filling manner (for a total of $16$ generations). Finally, a patient-specific computational lung model is constructed to predict regional lung mechanics, such as the strain or regional ventilation of the lung parenchyma during mechanical ventilation.
• Figure 4: Strain predictions for patient 2 during pressure-controlled ventilation at different points in time, showcasing the heterogeneous strain distribution within the lungs.
• Figure 5: Illustration of one exemplarily time-window underlying the validation, comprising 10 respiratory cycles. We compare the recorded flow rates and resulting tidal volumes exported from the mechanical ventilator (measured, clinical data) against the predictions obtained from the computational model. The waveforms are nearly identical.