A Damaged-Informed Lung Ventilator Model for Ventilator Waveforms

Abstract

Motivated by a desire to understand pulmonary physiology, scientists have developed physiological lung models of varying complexity. However, pathophysiology and interactions between human lungs and ventilators, e.g., ventilator-induced lung injury (VILI), present challenges for modeling efforts. This is because the real-world pressure and volume signals may be too complex for simple models to capture, and while complex models tend not to be estimable with clinical data, limiting clinical utility. To address this gap, in this manuscript we developed a new damaged-informed lung ventilator (DILV) model. This approach relies on mathematizing ventilator pressure and volume waveforms, including lung physiology, mechanical ventilation, and their interaction. The model begins with nominal waveforms and adds limited, clinically relevant, hypothesis-driven features to the waveform corresponding to pulmonary pathophysiology, patient-ventilator interaction, and ventilator settings. The DILV model parameters uniquely and reliably recapitulate these features while having enough flexibility to reproduce commonly observed variability in clinical (human) and laboratory (mouse) waveform data. We evaluate the proof-in-principle capabilities of our modeling approach by estimating 399 breaths collected for differently damaged lungs for tightly controlled measurements in mice and uncontrolled human intensive care unit data in the absence and presence of ventilator dyssynchrony. The cumulative value of mean squares error for the DILV model is, on average, ≈12 times less than the single compartment lung model for all the waveforms considered. Moreover, changes in the estimated parameters correctly correlate with known measures of lung physiology, including lung compliance as a baseline evaluation. Our long-term goal is to use the DILV model for clinical monitoring and research studies by providing high fidelity estimates of lung state and sources of VILI with an end goal of improving management of VILI and acute respiratory distress syndrome.

Pulmonary Fibrosis as a Result of Acute Lung Inflammation: Molecular Mechanisms, Relevant In Vivo Models, Prognostic and Therapeutic Approaches

“…The most significant contributor to ARDS morbidity and mortality in recent years was the COVID-19 pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), so named because of its high homology with SARS-CoV-1, the virus responsible for the outbreak of severe acute respiratory syndrome in 2002–2003 [ 36 , 37 , 38 ]. However, a great number of stimuli and diseases may serve as etiological factors of ALI and ARDS, including bacterial (Streptococcus pneumonia or Staphylococcus aureus [ 39 , 40 ]) and viral (influenza A virus or rhinovirus [ 41 , 42 ]) pneumonia, continuous mechanical ventilation [ 43 , 44 , 45 ], chemicals (chlorine, phosgene and industrial aerosols [ 46 , 47 , 48 ]), electronic cigarettes, and vape [ 49 , 50 ], acute brain injury [ 51 , 52 ], sepsis [ 53 , 54 ], acute pancreatitis [ 55 ] and many other pathologies.…”

Hypothesis-driven modeling of the human lung–ventilator system: A characterization tool for Acute Respiratory Distress Syndrome research

“…Practical applications, therefore, require analyzing characterization within the full patient-data environment, including knowledge of ventilator settings which the experiments here did not consider. Investigation of the relationship between lung physiology and estimated parameters can proceed in tandem with LVS waveform research, such as those of ventilator control mechanisms e.g., [48] , as well as models with explicit process mechanisms [26] and injury-specific waveform representation [37] . Although such considerations are beyond the scope of this discussion, the hypothesis-informed parameter descriptions can augment those analyses as low-dimensional representations of waveform data by providing both flexibility and interpretability in a discrete form.…”

“…The presented model determines a parametric representation of LVS waveforms using a method conceptually recent related to work recent [37] . Rather than use physiological relationships to model data generation, the proposed method simulates parametric waveforms, including dyssynchronous breaths, within an inferential scheme.…”

Computational lung modelling in respiratory medicine

“…Recently, machine learning (ML) techniques have proven helpful in enhancing the translation of computational lung modelling to the clinic [ 159 , 160 ]. Although ML often is directly applied in lung imaging to study clinical aspects of lung function such as ventilator parameter optimization [ 161 , 162 ], ML toolsets can also be used to facilitate several steps in common image-based modelling pipelines, that would otherwise require time-intensive processes. These steps include segmentation of the lungs and airways [ 163 – 165 ], isolation and segmentation of diseased regions such as inflammation, obtaining strains and displacements from image registration, and so on.…”