Efficient heat transfer fluids are essential for contemporary thermal engineering, as traditional liquids insufficiently deliver the required cooling and heating efficacy. Hybrid and ternary nanofluids, owing to their superior thermo-physical properties, offer a promising alternative. However, their nonlinear rheological behavior and complex transport mechanisms pose substantial modeling difficulties. To address these difficulties, we proposed an artificial neural network (ANN)-based numerical approach to investigate the parameters influencing the heat transfer of a magnetized Casson-based \(TiO_2 - MWCNT\) /ethylene glycol–water hybrid nanofluid and \(TiO_2 - MWCNT - Al_2O_3\) /ethylene glycol–water ternary nanofluid over a heated slender needle. We describe a new hybrid methodology that utilizes ANN and a modified Bvp4c approach to boost numerical modeling capabilities. We utilized similarity transformation to turn the governing partial differential equations into a set of ordinary differential equations. We use the MATLAB-based Bvp4c method to find numerical solutions to these equations. A parametric analysis is conducted to thoroughly examine the influence of essential dimensionless factors, including the nanoparticle shape factor m, the needle parameter c, the Casson parameter \(\beta\) , the magnetic parameter M, the radiation parameter Rd, and the thermophoresis parameter Nt. Quantitative analysis shows that the ternary nanofluid enhances the wall shear stress with a better skin friction coefficient than the hybrid nanofluid, though the local Nusselt number slightly decreases in the ternary nanofluid. Also, higher magnetic and Casson parameters cause higher flow resistance and reduced heat transfer, and higher thermal radiation improves temperature distribution. The ANN model shows a wide range of compatibility with numerical results and is a good alternative to an immediate thermal management system.
