Lung digital twin COVID-19 infection through multiphysics and multiscale HPC-modelling
Please login to view abstract download link
In response to the COVID-19 pandemic, researchers are using modeling to tackle the challenges of controlling global crises like epidemics. Most notable epidemiological models are SIR models and its derivatives, or molecular dynamics models of the binding of proteins. However, these models lack a deep understanding of viral evolution for developing effective therapeutic strategies. This study introduces a multiscale, multicellular, spatiotemporal model simulating SARS-CoV-2 infection in lung tissue. Bridging cellular to organ levels, this model aims to discover personalised therapeutic targets and simulate their impact on full-sized lung organs. Our approach integrates two distinct simulators, Alya and PhysiBoSS, with the ultimate objective of providing optimised, patient-specific interventions. Alya, a multiphysics and multiscale tool that simulates organs in HPC, models the airflow in the respiratory airways, tracking the transport of the virus to the alveoli. At the same time PhysiBoSS, an agent-based tool for multiscale simulations of cell populations and their environment interactions, simulates oxygen response, virus diffusion, and their impact on cells, assessing the alveolar state. The iterative coupling of these processes enables the prediction of a specific patient's evolution. At the organ level, a 3-dimensional geometry of the entire respiratory airways, based on real patient-specific clinical images, is employed. This in-silico digital twin model accurately reproduces various infection states, allowing the evaluation of viral deposition maps in the alveolar tissue. The resulting outcome parameters from these simulations are then adapted as inputs for PhysiBoSS to address the cell-level problem in the epithelium. Currently, the alveoli epithelium is the boundary of the Alya model, a surface embedded in 3D, while the standard input for PhysiBoss is a planar epithelium surface. In this work, we present a preliminary coupling between Alya and Physiboss models, that transforms the mapping between the epithelium 3D surface to a planar epithelium surface valid for the PhysiBoss simulation, transferring the solution from the first solver to the second. We hereby present the results of this transformation and some corner cases where the continuity of solutions is not maintained. In future works, we will adapt PhysiBoSS to simulate the 3D alveoli architecture so that this transformation is not needed and, thus, these corner cases will not be present.
