Overview

The MIMIC II dataset is a well known dataset comprising of many physiological signals and electronic health record variables. It is hosted by PhysioNet, and is a very helpful resource. It can be used for research into respiratory rate algorithms by extracting the relevant variables.

Details

Of particular relevance to this project, the dataset contains electrocardiogram (ECG), pulse oximetry (photoplethysmogram, PPG) and impedance pneumography (IP) signals acquired from critical care patients. It can be accessed here.

A Matlab script is available for extraction of this dataset as part of the toolbox of algorithms here. This script downloads the data and converts it into the necessary format for use with the algorithms used in this project.

Example

MIMICII_Signals
Exemplary MIMIC-II Signals: The three types of relevant simultaneous signals contained within the MIMIC-II Dataset are shown (from top): Impedance Pneumography (IP) signals; ECG signals; PPG signals.

Publications

Both this dataset and MIMIC I have been widely used in the literature to evaluate algorithms for estimation of respiratory rate. It was first used with the toolbox of algorithms presented in this project in this publication. This is a particularly helpful introduction to the use of MIMIC II in the field of respiratory rate estimation. We have also used the dataset, although not the toolbox, in this publication.

Link to download

The dataset can be accessed here. The Matlab script for extraction of the dataset can be downloaded here.