Search for a command to run...
This dataset presents an available collection of single‑channel ECG recordings obtained from volunteers using Apple Watch devices. Smartwatches have become an important source of biomedical data outside clinical environments, but researchers still lack open datasets that reflect the real conditions of ECG recording on wearable devices. Existing public datasets are mostly collected with medical‑grade equipment and do not include the noise, variability, and signal distortions typical for smartwatches. This limits the development and reproducibility of algorithms for ECG‑based biometric identification. The dataset introduced in this work helps address this problem. The dataset includes 30‑second ECG recordings from 39 volunteers, along with anonymized demographic information. Each ECG was exported from the Apple Health application as a PDF file. A modular Python pipeline was developed to extract, digitize, and analyze the signals. The pipeline consists of several stages: extracting ECG strips from PDF reports, stitching them into a continuous waveform, converting the graphical ECG into a digital time series using image‑processing methods, and applying filtering. The Savitzky–Golay filter was used to preserve the morphology of P, QRS, and T waves, which is essential for biometric identification. After digitization, the system automatically calculates key ECG parameters such as RR, PR, QRS, QT intervals, ST segment amplitude, and heart rate. The article also presents statistical analyses and visualizations of these parameters across different age and sex groups, showing the variability and characteristics of smartwatch‑based ECG signals.