Digital Twin Development for Human Lungs by Leveraging Reduced-Order Model Utilizing Star-Delta Network Transformation for Therapy of Patients With Acute Respiratory Distress Syndrome (Coronavirus Disease)

Abstract Acute respiratory failure caused by infection of the lungs leading to respiratory failure is termed acute respiratory distress syndrome (ARDS). It includes the recent pandemic, causing coronavirus disease (COVID-19). Higher mortality rates are commonly associated with critical care patients having ARDS. The treatment normally requires the use of mechanical ventilation, which potentially can lead to complications and increased lung injury termed as ventilator-induced lung injury (VILI). It is primarily caused by high inspiratory pressure or cyclical opening during respiration due to cyclic collapse during expiration. Heterogeneous ventilation is documented as the third mechanism responsible for VILI. To enumerate these mechanisms leading to VILI, a digital twin (DT) model was developed and implemented to study the flowrates and the pressure distribution in a simplified symmetric triple bifurcation human tracheobronchial airway model. A flow resistance network was generated using lumped parameter modeling techniques prior to formulating the digital twin. The individual resistance values for each zone in the bifurcating network were computed using flow parameters obtained from computational fluid dynamics (CFD) simulations. Five different inlet flows corresponding to Reynolds numbers (Re) varying from 100 to 2000 were simulated to generate points for the lumped parameter model. The goal of this study was to create a digital twin for the human respiratory system. The aim of this study was to investigate the variety of flow features induced in the airways for optimizing the settings during mechanical ventilation. These investigations focused on the pulmonary vasculature for developing a Lung Physiome. The main objective was to develop a flow-impedance network that was distilled from CFD simulations performed during this study. This was realized by utilizing star-delta network transformation (i.e., using an electrical network analogy), i.e., using a reduced-order model (ROM) based on results obtained from the CFD simulations of specific airway geometries. The proposed methodology was designed to simplify the numerical and computational procedures for obtaining the desired flow parameter predictions for the complex flow fields inside lung airways. These endeavors enabled the development of a simplified digital twin model based on flow resistance network predictions that align with anatomically consistent observations for the human respiratory system. Hence, this study helped demonstrate the feasibility of the development of predictive models that are required for optimizing the operating parameters of mechanical ventilators for individualized therapeutics of patients with ARDS and similar lung diseases (e.g., COVID-19).