NFS2ECG: Estimation of Electrocardiogram Signal Waveforms from Near-Field Sensing on Chest Wall

In this study, we propose a system to estimate electrocardiogram (ECG) signal waveforms from chest wall displacement recorded by a wearable electromagnetic near-field sensor (NFS) module with a small antenna. The NFS module can record the change in the input impedance of the antenna as a time-series NFS signal based on the variations in the near-field between the wall and the antenna, with minimal restrictions. Subsequently, the system generates a model to estimate the ECG waveforms from the NFS signals by focusing on the relationship between the signals. To evaluate the capability of the proposed system, we conducted an experiment to simultaneously record NFS and ECG signals from 12 participants. We compared frequency analysis-based and deep learning-based models to estimate ECG waveforms from the recorded NFS signals. The results indicated that our proposed system was capable of estimating ECG signal waveforms with a root mean square error 0.057±0.025 in the normalized amplitude using the frequency analysis-based model.