Neural network design for non-Newtonian Fe3O4–blood nanofluid flow modulated by electroosmosis and peristalsis

In this study, we present a novel approach that utilizes the Levenberg–Marquardt algorithm (LMA) based on artificial neural networks (ANNs) to evaluate the flow characteristics of a thermally evolved blood-based nanofluid in the presence of peristalsis and electroosmosis. The Casson fluid model is employed to govern the non-Newtonian characteristics observed in the flow of blood. In addition, the thermal properties of the nanofluidic medium in contact with platelet magnetite nanomaterials are also studied in detail. Further, the effects of thermal radiation, thermal buoyancy force, magnetic field and Joule heating are also given due consideration. The mathematically formulated two-dimensional equations describing the flow of Casson liquid are brought into their dimensionless form under the lubrication theory. A dataset for the proposed ANN models is generated to explore various scenarios of the fluidic model by varying the pertinent parameters using NDSolve in Mathematica. The computational approach utilizing LMA is deployed across three distinct phases of performance assessment, distributing the data into training, testing and validation sets at the proportions of 80%, 10% and 10%, respectively. This implementation involves the utilization of 10 hidden neurons. The utilization of regression analysis for testing, mean-squared error calculation, error histograms and correlation assessment in numerical replications of the ANNs is also examined to verify their capability, accuracy, validity and effectiveness. This study is crucial for understanding the peristaltic blood transportation in small blood vessels of living organisms.