MIMIC Dataset

Table of contents

  1. Overview
  2. Subsets
    1. AF Subset
    2. Ethnicity Subset
    3. Adult & Neonate Subset
  3. Importing the data into MATLAB
    1. AF Subset
    2. Ethnicity Subset
    3. Adult & Neonate Subset

Overview

Item Details
Links Dataset, Publication
Signals PPG, ECG, resp, BP, others
No. Subjs 1,000s
Protocol Recordings during critical care admissions

Subsets

We are currently developing approaches to extract the following subsets of the MIMIC dataset:

AF Subset

This subset contains ECG and PPG recordings of 1-hour duration, some of which were acquired during atrial fibrillation (AF), and the rest were acquired during normal sinus rhythm.

Item Details
Links TBC
Signals PPG, ECG, resp
No. Subjs 34 (18 in AF, 16 not in AF)
Protocol 34 critically-ill adults during routine clinical care. Data were measured using a bedside monitor at 125 Hz. Manual labels of AF and non-AF subjects were obtained from here, as described in [1]. Data were extracted from the MIMIC-III Waveform Database Matched Subset.

Ethnicity Subset

This subset contains ECG and PPG recordings of 10-minute duration, some of which were acquired from Black adults, and some of which were acquired from White adults.

Item Details
Links TBC
Signals PPG, ECG, resp
No. Subjs 100 (50 Black, 50 White)
Protocol 100 critically-ill adults during routine clinical care. Data were measured using a bedside monitor at 125 Hz. Data were extracted from the MIMIC-III Waveform Database.

Adult & Neonate Subset

This subset contains ECG and PPG recordings of 10-minute duration, some of which were acquired from adults, and some of which were acquired neonates.

Item Details
Links TBC
Signals PPG, ECG, resp
No. Subjs 100 (50 adults, 50 neonates)
Protocol 100 critically-ill patients during routine clinical care. Data were measured using a bedside monitor at 125 Hz. Data were extracted from the MIMIC-III Waveform Database.

Importing the data into MATLAB

I took the following steps to import each Subset into MATLAB:

AF Subset

  1. Download the MATLAB script for importing the data from here.
  2. Modify the MATLAB script by inserting the up.paths.root_folder and up.paths.save_folder into the setup_up function.
    • Note: The duration of recordings provided by the script can be adjusted with the up.settings.req_durn variable within the setup_up function.
  3. Run the MATLAB script to download the required files from PhysioNet, import the data into MATLAB, and collate the data into a single MATLAB data file, ready for analysis.

Ethnicity Subset

  1. Download the MATLAB script for importing the data from TBC.
  2. Modify the MATLAB script by inserting the up.paths.root_folder and up.paths.save_folder into the setup_up function.
    • Note: The duration of recordings provided by the script can be adjusted with the up.settings.req_durn variable within the setup_up function.
    • Note: The number of subjects in each group can be adjusted with the up.settings.no_subjs_per_ethnicity variable within the setup_up function.
  3. Run the MATLAB script to download the required files from PhysioNet, import the data into MATLAB, and collate the data into a single MATLAB data file, ready for analysis.

Adult & Neonate Subset

  1. Download the MATLAB script for importing the data from TBC.
  2. Modify the MATLAB script by inserting the up.paths.root_folder and up.paths.save_folder into the setup_up function.
    • Note: The duration of recordings provided by the script can be adjusted with the up.settings.req_durn variable within the setup_up function.
    • Note: The number of subjects in each group can be adjusted with the up.settings.no_subjs_per_group variable within the setup_up function.
  3. Run the MATLAB script to download the required files from PhysioNet, import the data into MATLAB, and collate the data into a single MATLAB data file, ready for analysis.

Copyright © 2021 Peter Charlton.