Thrombosis underlies the majority of complications in cardiovascular disease. Despite its high prevalence, there are few clinical tools that offer a personalized and comprehensive prognosis for thrombosis risk in cardiovascular disease patients. Existing thrombosis models focus on specific sub-processes of clot formation, but no single model has accounted for patient-specific arterial hemodynamics alongside biophysical and biochemical aspects. This is mainly due to the computational cost of scaling current approaches to the arterial vasculature, limiting their clinical use. We propose a computational model to simulate critical stages of arterial thrombosis, including platelet transport, adhesion, activation, and aggregation. Our model employs a multiscale approach treating blood flow as a continuum and platelets as discrete particles. Biophysical processes are represented by differential equations that describe ligand-receptor adhesion and intracellular calcium accumulation. We also introduce a time-decoupled representation of platelet biophysics to minimize computational cost. Our model requires information on the 3D anatomy, composition of prothrombotic surfaces, and blood composition (platelet count, receptor densities, and hematocrit). The model works as follows: 1) obtain flow fields from CFD simulations, 2) compute platelet trajectories using a physics-informed seeding strategy, 3) determine which platelets adhere initially, 4) generate a map of adhered and activated platelets, and 5) integrate platelet trajectory and activation state data to update platelet aggregation map. To test our framework, we conducted simulations in 3D geometries and flows representative of coronary arteries. We studied platelet adhesion, activation and aggregation for $Re=300$ flows in straight stenosed vessels ($50,70\%$) and curved vessels with $De\in[3,,36,57,82]$. Our results align with previous research suggesting that shear-mediated platelet GPIb-vWF bonds are critical for initial platelet adhesion in arteries, which facilitates stable GPVI-collagen bonds. Our simulations show platelet activation only in geometries where stenosis exceeds $50\%$. We observe that $De$ substantially alters the position of adhered platelets, but in the absence of stenosis, platelets do not activate. Next steps include data-driven compute acceleration to enable two-way coupling of the growing platelet aggregate to the CFD solver, as well as incorporating patient-specific platelet data.