Supporting data and code for device-aware machine learning of ECG and blood-pressure waveforms in cardiovascular aging

This research compendium contains participant-level derived data, numerical source data for Tables 1–4, Figures 1–6 and Supplementary Figure S1, trained ECG-only and combined ECG–blood-pressure models, analysis code, and publication-ready figures supporting the manuscript “Device-Aware Machine Learning Identifies Blood-Pressure Waveform Signatures of Cardiovascular Aging Beyond Single-Lead ECG.” The analysis derives from Autonomic Aging: A dataset to quantify changes of cardiovascular autonomic function during healthy aging, PhysioNet version 1.0.0 (doi:10.13026/2hsy-t491; RRID: SCR_007345). The original WFDB waveforms are not redistributed. The source metadata contain 1,121 records; 1,120 have waveform data because 0400.dat is absent from version 1.0.0. Age was released in intervals, and age_midpoint is an interval midpoint rather than an individual’s exact chronological age. Device 0 denotes the Task Force Monitor; Device 1 denotes CNAP 500 with MP150 acquisition. These models characterize healthy-aging-related physiological variation and are research tools, not diagnostic medical devices.