tranSMART timeseries demo

Install script for creating a docker based tranSMART installation on an ubuntu server. (tested with 18.04 LTS) The docker setup is available at https://gitlab.gwdg.de/medinfpub/tm_umg and will be downloaded by the install script.

A custom tranSMART version with time series visualization is available in the releases of this repository and will be installed. A transformed version of the MIMIC-III dataset (https://doi.org/10.13026/C2HM2Q) will be loaded automatically.

Requirements

The required programs of the install script will NOT be installed automatically. To install and run the tranSMART version you need the following applications installed on your machine:

• git
• docker (https://docs.docker.com/get-docker/)
• docker-compose (https://docs.docker.com/compose/install/)

Running the script

1. Download the latest release, which includes all required resources (https://github.com/JannikNickel/BA_transmart_timeseries/releases)
2. Navigate into the repository directory
3. Run the install script: ./install.sh YOUR_SERVER_URL
4. This will import a reduced MIMIC-III DEMO dataset with 4 time series variables. To import the full dataset use the command ./install.sh YOUR_SERVER_URL FULL The import of the reduced dataset takes 10-15 minutes. The import of the full dataset can take up to 24 hours or more depending on the server performance.

Testing the visualization

1. Open your server URL in a browser
2. The default login data is admin/admin
3. Create a subset of your choice
4. In the SmartR category, you can find the workflow 'timeseries'
5. Drag and drop high dimensional variables in the workflow (Available under Public Studies/MIMIC3_MIN/Subjects/Hospital Stays/Visit 01/Medical/Observations/)
6. Click fetch data to collect the variables from the database
7. In the Run Analysis tab, you can visualize the variables