Computational investigation and machine learning prediction of nanoparticle effects on hemodynamics in patient-specific aortic aneurysms.

This study presents a novel computational framework for nanoparticle-assisted drug delivery in abdominal aortic aneurysm (AAA) by characterizing patient-specific hemodynamics under realistic physiological conditions with integrating an artificial neural network (ANN). The model incorporates silver (Ag) nanoparticles within a fully transient, multiphase computational fluid dynamics (CFD) environment, where pulsatile blood flow is imposed using advanced user-defined function. Blood is modeled as a non-Newtonian fluid to capture its shear-dependent rheological behavior, enabling a more accurate representation of vascular flow dynamics. High-resolution patient-specific geometries are developed to investigate four distinct cases, including healthy and AAA arteries with varying geometrical complexities and multiple daughter vessels, both with and without nanoparticle infusion. The analysis focuses on key clinically relevant hemodynamic indicators, including velocity, normal and tangential wall shear stress (WSS). The results indicate that arterial geometry and blood rheology play a dominant role in governing flow behavior, with the most complex vascular configuration exhibiting the strongest hemodynamic response. Among the investigated cases, model 3 shows the highest velocity, decreasing by 52.94% compared to the healthy artery (model 1), while models 2 and 4 exhibit decreases of 82.35% and 54.90%, respectively at left iliac. The introduction of Ag nanoparticles enhances WSS across all models by 34.72%, 47.32%, 51.92%, and 39.67%, respectively, indicating improved flow modulation and more favorable conditions for drug transport and vascular interaction. Flow visualization reveals distinct regions of acceleration, recirculation, and nanoparticle accumulation, which are critical for understanding localized delivery mechanisms. The ANN confirming the robustness, accuracy, and predictive reliability of the developed models. Finally, the proposed framework establishes a clear relationship between arterial morphology, non-Newtonian blood behavior, and nanoparticle-driven hemodynamic modulation, offering valuable insights for optimizing patient-specific, nanoparticle-based therapeutic strategies in cardiovascular disease management.