Currently, the package supports PPG preprocessing and extraction of more than 400 features. The PPG pipeline was originally implemented for analysis of the AuroraBP database.
It provides:
- PPG preprocessing: Singal quality metrics, baseline extraction, etc.
- PPG feature extraction: time-domain, frequency-domain, statistical features ( >400 features)
- Compatibility with PPG recorded from 128 Hz to 500 Hz: tested with local devices and large datasets.
VitalPy is written in Python (3.9+). Navigate to the Python repository and install the required packages:
pip install -r requirements.txt
Import:
from src.ppg.PPGSignal import PPGSignal
Check the signal (make sure that the file waveform_df is in dataframe format and contains the columns 't' for time and 'ppg' for the signal values):
signal = PPGSignal(waveform_df, verbose=1)
signal.check_keypoints()
Get features:
signal = PPGSignal(waveform_df, verbose=0)
features = signal.extract_features()
Used file: measurements_oscillometric/o001/o001.initial.Sitting_arm_down.tsv
The following figure shows the mean template computed from all templates within the signal given as input.
All preprocessing steps are depicted. The final result should have filtered out all low quality waveforms.
Exemplary PPG keypoint extraction.
VitalPy is available under the General Public License v3.0.
If you use this repository or any of its components and/or our paper as part of your research, please cite the publication as follows:
A. Cisnalet al. "Robust Feature Selection for Continuous BP Estimation in Multiple Populations: Towards Cuffless Ambulatory BP Monitoring," IEEE J Biomed Health Inform, Under Review (2023).
@unpublished{vitalpy,
title={Robust Feature Selection for Continuous BP Estimation in Multiple Populations: Towards Cuffless Ambulatory BP Monitoring},
author={Cisnal, Ana and Li, Yanke and Fuchs, Bertram and Ejtehadi, Mehdi and Riener, Robert and Paez-Granados, Diego},
journal={IEEE Journal of Biomedical and Health Informatics},
year={2023},
note = "Under review"
}