Integrate Data-Driven Modeling and Multi-scale Measures Towards Tissue Function

Human tissues are highly organized structures with specific collagen matrix arrangements and the resulting mechanical properties varying from point to point. The effects of such heterogeneity play an important role for tissue function and failure. However, fundamental challenges present in understanding

how heterogeneity affects the growth and remodeling of various tissues and the related mechanical performance of these tissues, due to limited knowledge on understanding of the multi-scale interactions between cell signaling and heterogeneous matrix remodeling and their roles in tissue function. The central theme of this proposal is to develop a novel bio-mathematical framework that dynamically

integrates physics-based and d ata-driven modeling approaches with experimental measurements, to capture cell-matrix interactions in tissue multi-scale behavior and function. We aim to improve the fundamental understanding of the underlying biological and mechanical implications o f matrix heterogeneity, and advance towards our ultimate goal of providing a digital twin model for in-vivo tissue

degeneration on aneurysm growth. The team, formed by three Pl/Co-ls with complementary expertise on scientific machine learning, experimental tissue biomechanics, and biological tissue modeling, plans to ad vance on theoretical, computational, and experimental as pects towards the knowledge of fi broblast-medi ated extracellular matrix (ECM) remodeling via three specific aims: (i) A multi-scale

data-driven model will be developed to concu rrently couple our intra-cellular signaling network model, a novel agent-based cell population model, and a peridynamics model of tissue mechanics; (ii) E nriched by advanced multi-task/multi-modal operator learning techniques, an information-theory based model

adaptation-experimental design integration pipeline will be developed; and (iii) New quantitative knowledge on long-term fibroblast-mediated ECM remodeling will be obtained and validated through both in-vitro and ex-vivo experiments. As a result, our combined in-vitro/ex-vivo experiments and multi-scale

model will help identify for the first time how heterogeneity affects tissue remodeling and the related mechanical behaviors of tissue. The identified key features and optimized model forms enable the learning from small data regime, providing a step-stone towards the digital twin modeling f or in-vivo

aneurysm growth. The proposed model will have significant impacts on transforming our understanding of the roles of cell functi ons for determining tissue structures/mechanics, and thereby enabling model-based virtual screening to identify and optimize therapeutic interventions. Our multi-scale model will of fer a

powerful prediction tool to the multi-scale biological mechanisms in many other bio-tissue systems, such as blast-induced traumatic brain injury, aging of cartilage tissue, and bone reconstruction.