The progression of arterial stenosis disrupts hemodynamic behavior and thermal regulation, contributing to cardiovascular complications. This study develops a computational model for unsteady blood flow through a time-dependent, multi-stenosed artery, incorporating viscoelastic effects via the Jeffrey fluid model and memory-dependent characteristics using the Caputo–Fabrizio fractional derivative. The model accounts for biophysical influences including magnetic fields, thermal radiation, body acceleration, chemical reactions, and the Soret effect. A ternary hybrid nanofluid composed of gold \(\left( {{\text{Au}}} \right)\) , silver \(\left( {{\text{Ag}}} \right)\) , and Multi-walled carbon nanotubes \(\left( {{\text{MWCNT}}} \right)\) is employed to enhance heat and mass transfer properties. The governing equations for velocity, temperature, and concentration are addressed using a semi-analytical approach that involves Laplace and finite Hankel transforms, combined with numerical integration. To estimate key transport quantities like skin friction, Nusselt number, and Sherwood number, an Artificial Neural Network (ANN) trained via the Levenberg–Marquardt algorithm is incorporated. The model’s predictive capacity is evaluated under varying physiological parameters using different data partitioning strategies, with the best performance observed in the 70%–15%–15% split. The results indicate that higher fractional order and magnetic field strength reduce flow velocity, while increased Darcy and Reynolds numbers enhance it. Thermal radiation and the Soret effect positively influence thermal and solute transport, respectively. The ternary hybrid nanofluid leads to a 36.75% improvement in heat transfer and a 35.32% enhancement in mass transfer. This integrated modeling framework offers a comprehensive approach for analyzing nonlinear hemodynamic systems, with potential applications in biomedical procedures such as hyperthermia and drug delivery.
