Skip to main content

Pneumonia Prediction Step 2 - Initial Project Creation & Data Extraction

Creating a Project in the FCP UI

Dr Emily Smith.png

Dr. Smith is now officially ready to start her pneumonia prediction project. She has decided to perform the majority of her work using the Rhino SDK. To get started, the first thing she will need to do is create a Project using the Rhino UI.

  1. Log in to the FCP UI, entering your provided username and password. If this is your first time logging in, you will be required to change your initial password and accept the terms of the EULA.
  2. Create a new Project by clicking on the Add New Project button in the top-right corner of the dashboard.
  3. Enter the following information in the modal window that appears:
    • Name: Federated Datasets and Predictive Modeling
    • Description: Federated Datasets and Predictive Modeling
    • Permission Policy: Expand this section to explore the various configurable permission policies and personas that are available to you. Follow this link for more information about permissions. For this tutorial, the default permission policy is acceptable.
  4. Click the Create Project button to create your project. Once clicked, you should see your newly created project within the Projects Dashboard.
    • Click on the stacked 3-dot menu within the top-right corner of your new project's card.  A menu will appear with the option to Copy UID.  Click that option and save your project's UID for later; you will need it when you begin working within the associated Jupyter notebook.  Follow this link for more information about retrieving your Project's UID.

Add your Collaborator

Dr Mark Jones.png

To federate the project, Dr. Smith must now add her collaborator, Dr. Mark Jones.  Dr. Jones has agreed to make data available for Dr. Smith to use in her pneumonia prediction project.  For her to be able to use the data, she must first create a collaboration request within her newly created project in the Rhino FCP.  When Dr. Smith initiates the collaboration request, an email will be sent to Dr. Jones.  Dr. Jones will need to review the Project's Permissions Policy. If they are acceptable, he will accept the collaboration, and Dr. Smith will be able to access resources Dr. Jones imports within the Project. To view a walk-through of Adding a Collaborator, follow this link.

The Collaboration.png

  1. Click Collaborators from the left-hand navigation
  2. Once on the Collaborators page, click the Add New Collaborator button in the top-right corner. If the button is disabled, that means your user is not denoted as a Workgroup Admin.  If you encounter this issue, please contact support@rhinohealth.com
  3. Within the Add New Collaborator modal, search or scroll through the list of potential Collaborators until you find Rhino Sandbox Health System and select the checkbox next to it. 
  4. Check the box I approve inviting selected collaborators to this project
  5. Press the button Add Collaborator to initiate the collaboration invitation process.
  6. You will receive approval within a few minutes. Congratulations! You are now able to use datasets from the Rhino Sandbox Health System in addition to the datasets available in your Rhino Client.

Sandbox Meta Note: We simulate having multiple datasets from two different organizations, also called the “cross-silo” setting in FL, by splitting the original MIMIC-IV dataset into multiple partitions. Each partition will represent the data from each organization. This is done purely for experimentation purposes. In real-world situations, there’s no need to partition data because each organization already has its own data, so the data is naturally partitioned.

 

Data Extraction

Armed with SQL, Dr. Smith delves into her hospital's Electronic Health Records (EHR) system, identifying patient records that match her criteria. With data in hand, she's ready to import the textual data into the FCP as a Dataset.  After importing her Dataset of text-based patient data, Dr. Smith will gather each previously identified patient's chest x-rays (CXR) from her hospital's Picture Archiving and Communication System (PACS) system.  Once she has identified all the CXRs, she will import the DICOM files as another Dataset with a lookup table to map back to the original EHR data.

Help Dr. Smith extract her data from both the EHR and PACS system and create Dataset in the FCP by completing Notebook #1: Data Extraction

 

Supplemental Resources

 

 

If you need support at any time, feel free to contact support@rhinohealth.com.

 

Continue to Pneumonia Prediction Step 3 →

Was this article helpful?