Wideband Circularly Polarized Conformal Antenna with Physics-Informed Neural Network Modeling for IoBNT Capsule Endoscopy.

The convergence of artificial intelligence, biotechnology, and the Internet of Bio-Nano Things (IoBNT) is enabling the creation of a new generation of intelligent in-body medical devices for continuous diagnosis and monitoring. In this context, a compact, wideband, circularly polarized conformal microstrip antenna is proposed for capsule endoscopy applications. The antenna is integrated along the inner wall of a 10 mm-diameter capsule and achieves an impedance bandwidth of 2.06-5.39 GHz (89.39%), maintaining stable matching under varying biological tissue conditions. A 3 dB axial ratio bandwidth (ARBW) of 2.31-3.14 GHz (30.45%) ensures reliable circular polarization and robust wireless communication in lossy and dynamic in-body environments. To extend beyond conventional electromagnetic analysis, a physics-informed neural network (PINN) framework is introduced to model the thermal response of biological tissues based on the governing bioheat equation. This AI-driven approach enables fast and generalizable prediction of temperature rise under varying operational conditions without repeated numerical simulations. At 2.45 GHz, the antenna exhibits a maximum gain of -31.1 dBi with a radiation efficiency of approximately -34 dB, consistent with in-body propagation constraints. Simulation and experimental results in realistic tissue phantoms, including muscle, small intestine, large intestine, and stomach, confirm stable wideband and polarization performance. Specific absorption rate (SAR) analysis demonstrates compliance with IEEE C95.1-2019 safety limits, while link budget evaluation validates reliable telemetry over a 1-3 m communication range. The integration of advanced antenna design with physics-informed machine learning provides a scalable framework for intelligent, safe, and adaptive IoBNT-enabled capsule endoscopy systems.