Reconstruction of ECG from Ballistocardiogram Using Generative Adversarial Networks with Attention

Abstract Electrocardiogram (ECG) is widely used to provide early warning signals for cardiovascular diseases. However, traditional twelve-lead ECG monitoring methods and smartwatch-based home solutions are unable to achieve daily long-term monitoring. Therefore, in this work, we propose a system to reconstruct ECG signals from non-contact Ballistocardiogram (BCG) signals. First, we synchronously collect BCG and ECG signals using fiber optic sensors and an ECG machine, and preprocess the signals to obtain a training set. We train the Att-SNGAN model using this training set to reconstruct ECG signals from BCG inputs. Experimental results show that the reconstructed ECG has a mean absolute error (MAE) of only 0.0651, and it holds promise for future applications in cardiac cycle monitoring and heart rate variability (HRV) analysis, demonstrating the system's effectiveness. This work provides new solutions for home ECG monitoring.