Classification of MIMIC-IV ECGs

[sanketahegde]

Prerequisites

• Put an X between the brackets on this line if you have done all of the following:
  • Checked the online documentation: https://mimic.mit.edu/
  • Checked that your issue isn't already addressed: https://github.com/MIT-LCP/mimic-code/issues?utf8=%E2%9C%93&q=

Description

Description of the issue, including:

• what you have tried
• references to similar issues
• queries demonstrating your question (if applicable)

I am currently working with MIMIC-IV ECGs as part of one of my research project. To continue using MIMIC-IV ECGs, I first have to classify the ECGs into different disease categories based on their diagnosis.

I see that there are multiple report columns in machine_measurements.csv file. I am currently unsure about how to separate the ECGs into healthy (normal sinus rhythm) and unhealthy or other disease conditions. Since, there are multiple report columns and each column have different diagnosis statements, this makes it difficult to get the normal ECGs.

Any suggestions or feedback is appreciated!!
Looking forward to get some answers.

Thank you!,
Sanketa

[sanketahegde]

@alistairewj

[briangow]

@sanketahegde , the data provided by the hospital had multiple report columns - we are preserving this structure in the table your are referring to.

I'm not a clinician but a conservative approach could be to look for a row that says "Sinus rhythm" and "Normal ECG" in any of the columns, without any other text / entries in the report columns.

[sanketahegde]

@briangow Thank you for the reply.
I was thinking of doing what you mentioned but getting a clarification from your side helps.

[TAUforPython]

@sanketahegde I have a same question. But I research MIMIC database (csv file), analyse report's columns myself. I've added new column 'Healthy_Status' manually. I've looked every row and mark all rows without record 'normal ECG' as a 1 in new column 'Healthy_Status'. So, all other rows I've marked automatically as a 0.
If You don't mind against such methodology, please be free and get my csv-dataset with 'Healthy_Status' from my repo.