Artificial neural network modeling of magnetic nanoparticle-enhanced Sisko blood nanofluid flow over an inclined stretching surface with non-uniform heating and thermophoretic effects

In the evolving field of fluid power and thermal systems, artificial neural networks (ANNs) are increasingly recognized for their robust ability to address nonlinear, coupled, and high-dimensional fluid dynamics problems. This study presents a neural network-assisted investigation of magneto-hydrodynamic Sisko nanofluid flow modelled as a blood-based magnetic suspension over an inclined stretching surface influenced by non-uniform heat generation and thermophoretic effects. The governing partial differential equations derived from mass, momentum, and energy conservation laws with complex boundary conditions are reduced to nonlinear ordinary differential equations through similarity transformations. The resulting system is first solved using MATLAB’s bvp4c solver, and the generated data is then used to train, validate, and test an ANN framework based on the Levenberg Marquardt backpropagation algorithm (BPLMA). The ANN model exhibits high predictive accuracy, with relative absolute errors ranging from 10⁻³ to 10⁻⁷ compared to the reference solution. The thermo-fluidic behaviour of shear-thinning and shear-thickening regimes is analysed under different concentrations of magnetic nanoparticles such as iron oxide and cobalt ferrite. For a 10 percent volume fraction increase, enhancements in heat transfer and reductions in mass transfer are observed, reaching up to 10 percent and 18.9 percent for iron oxide and 9.8 percent and 12 percent for cobalt ferrite, respectively, depending on the fluid rheology. Visualizations of streamlines, temperature fields, and concentration contours reveal intricate flow structures and nanoparticle distributions, offering valuable physical insights. Statistical evaluations including regression analysis, error histograms, and model fitness further support the reliability of the ANN approach. This work introduces a powerful hybrid computational methodology that integrates numerical simulation with machine learning to analyse non-Newtonian nanofluid behaviour and contributes to advancements in biomedical engineering, heat exchanger design, smart cooling systems, and microfluidic devices in fluid power applications. This work presents a novel computational framework that combines traditional numerical simulation with artificial intelligence to analyse complex non-Newtonian nanofluid behaviour. Unlike traditional methods that are often computationally intensive, the ANN model offers fast, accurate predictions and strong generalization across varying conditions. The novelty of this hybrid approach lies in its ability to enhance traditional techniques with AI driven efficiency, making it well suited for applications in biomedical engineering, heat exchanger design, smart cooling systems, and microfluidic devices.