Databases

• MIMIC

MIMIC

The MIMIC database, which stands for Medical Information Mart for Intensive Care, is a North American database containing data from over 50,000 patients admitted to intensive care. It is one of the most widely used intensive care databases, due to its free access.

Despite having data of imperfect quality, it serves as a good foundation for learning to manipulate data from health data warehouses (HDWs).

It exists in several versions, with the most recent being MIMIC-IV.

No, it is not necessarily required to have programming knowledge to manipulate the data in this database.

Handling the database itself will require programming knowledge, especially in SQL. It is a database with a particular schema, which is difficult to grasp.

However, it is possible to manipulate this data entirely with a graphical interface: this is one of the reasons why LinkR was created.

The MIMIC database includes test databases for versions III and IV, which contain anonymized data from 100 patients and are publicly accessible.

You can download the data here:

• MIMIC-III test: data with the MIMIC data schema
• MIMIC-IV OMOP test: data with the OMOP data schema

To access the complete databases, you need to complete a few steps.

Visit the MIMIC-III database page.

You will see this box at the bottom of the page:

You must start by registering on the site physionet.org.

You will need to submit an access request to Physionet, providing some information and the contact details of a supervisor or colleague, who will receive an email.

You will then need to complete the CITI Course, a necessary training to access data hosted on the Physionet site. The different steps are detailed here.

You can then download the certificate once the CITI Course is completed, and you can submit it here for validation by the Physionet team.

Finally, you will need to sign the data use agreement.