AI and Digital-Twin Synergy for Field Optimisation for Targeted Drug Delivery

Precise mapping of magnetic fields is crucial for magnetic drug targeting, microrobotics, and magnetically actuated biomedical devices. In this paper, we present an integrative approach that combines Finite Element Method Magnetics simulations, AI-generated field reconstructions, and OCR-assisted teslametry measurements. A neodymiumironboron permanent magnet block 30 × 20 × 20 mm3 was surveyed as a test specimen. An exact FEMM-determined magnetisation map was used to train two deep-learning surrogates: a Volumetric Network and a Physics-Informed Neural Network. The network results show a practically identical correspondence to those generated by FEMM, with a computational speedup of over fifty times. OCR measurements of the axial component Bz confirm agreement within experimental uncertainty. The FEMMAIOCR approach provides rapid, experimentally validated magnetic-field characterisation for biomagnetic applications.