Machine Learning-Based Matching Medium Design for Implant Communications

Matching media are used in various applications to increase the power transmitted into human body. The selection of optimum matching medium (MM) permittivity is not a straightforward task, as value maximizing depends on thickness MM and electromagnetic properties target tissue. In this article, computationally heavy empirical approach machine learning (ML)-based utilized for MM. demonstrates that can $|S_{21}|$ values up 8 dB, which validated with measurements. Next, an ML-based tool proposed predict any tissue thickness. A 1-D convolutional neural network followed by multilayer perceptron trained simulated average Poynting vector magnitudes (APVMs) muscle fat tissues. APVM dipole length given system parameters predicted artificial network. accuracy calculated comparison results analysis found be 1% 12.3% mean absolute percentage error APVM, respectively. decreases time required milliseconds.