This package implements the methods discussed in [1].
- Clone the repo.
- Open MATLAB script particle_filter/particle_filter.m.
- Modify lines 10-11 to set
record_nameas the path to your file. - Run the script. Currently, it'll print out the sensitivity and positive predictivity for the PF, GQRS, and WABP.
- The scripts that were used to prepare the MIT-BIH and MGH/MF databases and to generate statistics can be found in the scripts/ directory.
- The repo comes with setp1 and setp2 available at the Physionet 2014 challenge website.
- We assume that the signals come in a form compatible with the WFDB toolbox (we've included the toolbox in this repo).
- We utilized some of the SQI functions from Alistair Johnson's entry. The original source can be found here. Their paper detailing their work can be found here.
- We'll be updating the functions/scripts over time to improve documentation.
- We've provided the utility script that we used to generate our graphs.
[1] Hugh Chen, Yusuf B. Erol, Eric Shen, Stuart Russell, "Probabilistic Model-Based Approach for Heart Beat Detection", Physiological Measurement, Vol. 37, No. 9, August 2016, link.
If you have any questions, feel free to send an email to hugh.chen1{at}gmail.com and ericshen{at}berkeley.edu.