Data and code: Quality Assessment of Photoplethysmography Signals

Published: 11 June 2026| Version 1 | DOI: 10.17632/g64nf2pgbw.1
Contributor:
Rong Bian

Description

This dataset consisted of 18,763 10-second PPG signal segments, sourced from two publicly available datasets from PhysioNet: (1) the Brno University of Technology Smartphone PPG Database (BUT PPG) available at https://physionet.org/content/mimic3wdb-matched/1.0/ ; (2)the MIMIC-III Waveform Database Matched Subset available at https://physionet.org/content/butppg/2.0.0/. The BUT PPG Database was collected at a sampling rate of 30 Hz during both rest and physical activity. Conversely, the PPG signals in the MIMIC-III subset were originally sampled at 125 Hz, with varying segment lengths. To standardize the quality assessment, all signals were resampled to 30 Hz and segmented into non-overlapping 10-second clips. Each segment was assigned to a quality label of A, B, or C. In total, 18,763 PPG segments were obtained, consisting of 5,912 class A, 6,385 class B, and 6,466 class C segments. Here are part of the data.

Files

Institutions

Categories

Biomedical Engineering

Licence