ANN Prediction of Ellis Penta-Hybrid Nanofluid Transport in MHD, Nonlinear Radiation, Binary Reactions and Activation Energy Effects of Stenosed Artery

This study analyzes the two‑dimensional flow of a magnetized non‑Newtonian Ellis‑type penta‑hybrid nanofluid flowing through a stretching stenosed artery. The mathematical model incorporates velocity slip along with convective thermal and concentration boundary conditions to better represent physiological transport processes. Important physical effects such as nonlinear thermal radiation, viscous dissipation, Joule heating, porous medium resistance, and binary chemical reactions with activation energy are also included. To efficiently predict the transport behavior, an Artificial Neural Network (ANN) based on a feed‑forward multilayer perceptron trained using the Levenberg–Marquardt algorithm is developed, with the available data divided into training, validation, and testing sets in the ratio 70:15:15. The results indicate that the velocity decreases with increasing suction parameter S, magnetic parameter M, porous parameter Kp, and velocity slip parameter ω, while it increases with the stretching parameter ξ and curvature parameter γ. The temperature profile decreases with higher values of S and γ, but increases with the Eckert number Ec and nonlinear thermal radiation parameter Nr. The concentration distribution decreases as ξ, Schmidt number Sc, and chemical reaction parameter Rc increase. Moreover, the skin friction coefficient increases with S, ξ, Kp, and M. The Nusselt number slightly increases with γ and S, increases significantly with Nr, and decreases with Ec, whereas the Sherwood number increases with ξ, Sc, and Rc. The ANN predictions show excellent accuracy, with the best validation performance ranging from 4.4699 × 10⁻¹¹ to 2.5775 × 10⁻⁹ and regression values of R = 1 for all datasets. Prediction errors remain very small, mostly between −2.3 × 10⁻⁴ and 2.5 × 10⁻⁴, with deviations generally within ±5 × 10⁻⁴, confirming the high reliability of the ANN model. These findings demonstrate that ANN provides a fast and accurate surrogate framework for predicting velocity, temperature, and concentration fields in complex bio‑thermal transport systems. The outcomes of this study may be useful in biomedical engineering applications, particularly in improving the understanding of heat and mass transfer in stenosed arteries and in the design of advanced nanofluid‑based therapeutic or diagnostic technologies.