ANN validation and RSM optimization of Cattaneo-Christov heat flux in MHD stenotic artery trihybrid nanofluid flow

This study investigates shape-dependent thermal transport in a blood-based Cu–Al₂O₃–Au trihybrid nanofluid using a local throat-level similarity model motivated by stenosed arterial geometry. The model incorporates magnetohydrodynamics, thermal radiation, viscous dissipation, porous-medium resistance, Darcy–Forchheimer drag, and Cattaneo–Christov non-Fourier heat-flux. The key novelty lies in combining mixed-shape nanoparticle modeling with response surface methodology (RSM) optimization and artificial neural network (ANN) surrogate prediction for arterial-flow configurations relevant to thermal therapy. Shape-dependent property correlations compare a fully spherical trihybrid suspension with a mixed-shape case (spherical Cu/Al₂O₃, cylindrical Au). Using similarity transformations, the governing equations are converted into coupled nonlinear ODEs and solved numerically with a collocation-based boundary-value solver. Parametric analysis covers magnetic parameter (0 ≤ M ≤ 3), Eckert number (0 ≤ Ec ≤ 1.5), radiation parameter (0 ≤ Rd ≤ 0.75), Darcy parameter (1 ≤ K ≤ 5), thermal relaxation parameter (0 ≤ α t ≤ 0.1), and nanoparticle volume fraction (0.03 ≤ ϕ ≤ 0.1). RSM with a face-centered central composite design applied over the bounded optimization space for prediction of Nusselt number ( Nu ), followed by feedforward ANN, identifies the high- M , high-ϕ, and low- Rd region as favorable for local wall heat-transfer enhancement. The solver-confirmed constrained optimum ( M = 2.9935, ϕ = 0.0525, and Rd = 0.00012) yields Nu = 12.286, with ANN predicting Nu = 12.1721 (relative deviation: 0.93%). The mixed shape configuration increase mean velocity by 0.56–6.18% and mean temperature by 0.49–12.88% versus spherical particles. Overall, the combined numerical–RSM–ANN framework offers a reproducible route for screening design space and optimizing trihybrid nanofluid heat transfer in magnetically assisted arterial thermal-therapy modeling. Clinical trial number: Not applicable. This study did not involve a clinical trial.