Metastasis remains the leading cause of cancer-related mortality. To spread, cancer cells escape the primary tumor site, traverse heterogeneous tissues and the extracellular matrix, evade the immune system, and establish secondary tumors. Throughout this process, cells constantly interact with microenvironments that resist, relax, and remodel in response to force. These mechanical interactions shape how cells adhere, generate traction, maintain structural integrity, and adapt to their surroundings. This dissertation investigates how mechanics regulates cancer progression and immune dynamics across scales, combining theoretical modeling, computational analysis, and experimental validation.

At the single-cell scale, we examine how the time-dependent mechanics of the extracellular matrix modulate cell migration. By regulating cell-matrix adhesion lifetimes, matrix viscoelasticity alters the statistics of cancer cell motion and drives a transition from sub-diffusive to super-diffusive migration. This transition changes how cancer cells explore their microenvironment and provides a physical mechanism linking adhesive dynamics to invasive behavior. We then study how extracellular matrix mechanics modulates T cell activation at the immune synapse by showing how receptor engagement and integrin-mediated adhesion cooperate to regulate force generation, spreading, and downstream signaling.

At the multicellular scale, we model the tumor as a mechanically active continuum in which growth, transport, and stress generation produce spatially heterogeneous cellular states. Building on a mechano-electro-osmotic framework linking metabolic gradients and ion transport to osmotic swelling and residual solid stress, we develop AI-based image analysis tools to quantify these mechanical signatures across tumor systems. Using automated segmentation, we extract spatial patterns in cell morphology and nuclear deformation from in vitro spheroids, ex vivo tumor models, and in vivo tissues. This analysis supports a physical pathway linking tumor mechanics to localized DNA damage and genomic instability, while providing a quantitative framework that reveals principles of spatial organization and their governing mechanisms.

Finally, we extend this multiscale perspective to nuclear organization, where mechanical cues can reshape chromatin architecture and gene regulation. To study these genomic changes, we develop multimodal machine learning tools that integrate chromatin conformation, chromatin accessibility, and gene expression to learn shared representations of genome organization. Together, these studies demonstrate physical principles by which cells sense, transmit, and adapt to mechanical forces, revealing how mechanics contributes to migration, immune regulation, tumor evolution, chromatin organization, and therapeutic response.