" Chest\u2013Palm Paired ECG Dataset Combining Real and PTB-XL\u2013Derived Signals"

"This dataset is designed to support the development, training, and benchmarking of deep learning models for ECG signal denoising. It comprises paired clean and noisy ECG recordings obtained from both chest and palm acquisition setups, supplemented with augmented samples derived from the PTB-XL dataset. All signals are standardized to a uniform sampling frequency to ensure consistency across sources.The dataset is constructed through a structured pipeline that includes signal preprocessing, noise modeling, data augmentation, and rigorous validation procedures. A strict patient-wise splitting strategy is enforced to eliminate data leakage and ensure realistic generalization performance. The dataset contains carefully aligned input\u2013target pairs suitable for supervised learning tasks, enabling models to learn effective noise suppression mappings.To promote robust evaluation, the dataset is intended for use with multi-seed experimentation and aggregated performance reporting. It supports benchmarking across diverse model architectures, including proposed methods, ablation variants, and capacity-matched baselines.This dataset facilitates reproducible research by providing standardized data partitions and a consistent evaluation framework. It is particularly suited for advancing research in biomedical signal processing, real-world ECG denoising, and noise-robust cardiac monitoring systems, where reliable signal reconstruction is critical for downstream clinical and analytical applications."