Most modern cardiac monitoring devices, including smartwatches and portable ECG monitors, provide only a single-lead ECG signal, limiting their ability to offer a comprehensive cardiac analysis. However, standard clinical ECG systems rely on 12 leads to accurately diagnose heart conditions. To bridge this gap, we propose a TimeGAN-based neural network approach to generate full 12-lead ECG recordings using only single-lead inputs. TimeGAN is tailored for sequential data and combines a feature encoder, a signal restoration module, a temporal generator, and a discriminator specialized in time sequences to produce physiologically plausible and time-aligned ECG reconstructions. Our model was trained and evaluated on benchmark ECG datasets (PTB-XL, PTB, and INCART) and demonstrated superior performance compared to existing reconstruction methods. The results show a Pearson correlation of 0.75, significantly improving over traditional models. This research highlights the potential of data-driven generative models in enhancing remote cardiac monitoring, paving the way for more accurate AI-assisted diagnostics in wearable healthcare technology.