Implantable Battery Monitoring Using Hybrid Electromagnetic-Ultrasonic Transmission Method and CNN-LSTM Algorithm

Accurate, non-invasive State of Charge (SoC) monitoring for Implantable Medical Device (IMD) batteries is essential to prevent device failure, optimize replacement scheduling, and reduce patient risks from unnecessary surgeries. Traditional invasive methods consume battery power or lack accuracy, limiting their use in IMDs. This study aims to develop an innovative ultrasonic transmission method for reliable, non-invasive IMD battery monitoring. We propose a Hybrid Electromagnetic-Ultrasonic Transmission Method that uses externally generated 2 MHz, 0 dBm microwaves to wirelessly power piezoelectric transducers on the battery surface. These transducers convert microwave energy into ultrasonic waves, enabling transmission-mode SoC assessment through a shortened acoustic path. The system measures S21 transmission coefficient magnitude and phase, which are then processed using a CNN + LSTM model to predict SoC. Testing was conducted in simulated tissue environments across various battery types and charging conditions. The method achieved SoC estimation accuracy of approximately 95% within a 5% deviation across all tested conditions, achieving an MAE < 2% across datasets, including different battery models, charging rates, and battery health states. The system demonstrated consistent performance without requiring complex instrumentation or additional IMD battery consumption. The proposed method enables continuous, passive SoC monitoring using a simple external setup, optimizing clinical decisions for battery replacement timing and reducing unnecessary surgical interventions and healthcare costs.