Understanding how molecular signaling, cellular behavior, and biomechanical forces jointly regulate vascular inflammation remains a central challenge in atherosclerosis research [1]. Experimental evidence indicates that lipid dysregulation, immune activation, and mechanical loading interact through nonlinear feedback loops whose relative importance varies across individuals. Capturing this variability requires computational frameworks that integrate biological knowledge with biomechanical modeling while remaining interpretable and compatible with data-driven inference [2, 3, 4]. We present a graph-informed mechanochemical modeling framework that integrates experimental data and computational biomechanics for hypothesis testing and virtual-patient generation. The approach is based on a curated multiscale interaction graph incorporating lipid species, cytokines, macrophage and smooth muscle cell phenotypes, extracellular matrix turnover, oxidative and osteogenic cues, and mechanical drivers such as strain and shear stress. Nodes represent biological or biomechanical entities, while edges encode activating, inhibiting, or modulatory interactions. A rule-based grammar translates the graph into coupled mechanochemical models by assigning weighted Hill-type operators, defining cross-scale couplings, and specifying ordinary differential equation controllers for phenotype transitions. To support calibration and uncertainty-aware analysis, interactions are assigned low, medium, or high influence levels mapped to normalized kinetic coefficients, reducing parameter dimensionality while preserving interpretability. This enables automated generation of virtual patients and systematic exploration of competing mechanistic hypotheses using Bayesian simulation-based inference. The framework produces compact coupled ODE–PDE models capturing biochemical transport, cellular dynamics, and tissue mechanics, yielding distinct mechanistic signatures across virtual-patient populations. Overall, this work provides a scalable and interpretable workflow for integrating experimental data and computational models in vascular biomechanics, supporting robust prediction and personalized mechanochemical analysis. REFERENCES [1] Libby, P. (2021). Nature 592, 524–533. [2] Buehler, M. J. (2024). MLST 5(3), 035083. [3] Gierig, M., Gaziano, P., Wriggers, P., and Marino, M. (2024). J Biomech 166, 112058. [4] Marino, M., Sauty, B., and Vairo, G. (2024). Biomech Mod Mech 23, 1091-1120.