Predictive design using physics-informed neural network for the bioconvection of an electrically conducting casson hybrid nanofluid with heat and mass flux model

This is the authors' abstract. We don't add key points for this paper.

The development of intelligent computational models for complex bio-thermal transport systems is important for biomedical applications involving controlled heat and mass transfer. The objective of the present study is to develop a physics-informed neural network (PINN)-based predictive framework for analyzing the MHD stagnation-point bioconvective transport of an electrically conducting blood-based Casson hybrid nanofluid. The novelty of this study lies in the unified PINN treatment of MgO–SiO₂–blood hybrid nanofluid transport incorporating motile microorganisms, thermal radiation, viscous dissipation, chemical reaction, and Cattaneo–Christov heat and mass fluxes. The governing PDEs are transformed into a coupled system of nonlinear ODEs through similarity transformations, while the governing physical laws and boundary conditions are embedded directly into the PINN learning process to predict the velocity, temperature, concentration, and microorganism-density fields. The results demonstrate that the magnetic parameter enhances the velocity distribution under the considered stagnation-point flow conditions, whereas thermal radiation elevates the temperature field and chemical reaction suppresses species concentration. Quantitatively, at β = 3 and M = 0.1 , the skin-friction coefficient decreases from 0.64457 for the MgO–blood nanofluid to 0.40608 for the MgO–SiO₂–blood hybrid nanofluid, corresponding to an approximately 37.0% reduction . Furthermore, the hybrid nanofluid exhibits approximately 56.33% higher Nusselt number than the MgO–blood nanofluid under the corresponding investigated conditions, demonstrating enhanced heat-transfer performance. The low MSE, MAE, and loss values further demonstrate the predictive capability and stable convergence of the proposed PINN framework.