Project

A project is an organizational unit in the context of the FLAME platform, which represents the collaboration between different participants in regard to a specific research or analysis project. It contains an initial risk assessment as well as a high level description of the requested data.

In the Home section press Projects in the menu on the left-hand side. At first, you will lead to the outgoing-proposals section, where you can see a list of all the projected created by this realm. Via the searchbar you can filter by name or id. You have the option to adjust the already existing projects via the List-icon or to delete the project via the bin-icon. Furthermore, you can see, who exactly created that project.

Moreover, you have the option to switch to the incoming section or to create a new proposal on the left side between the menu and the overview-list.

Creation

After pressing the Create-button, it will take us to a new form. Here you can define principal aspects of the analysis. You can define ...

• ... a title of the project.
• ... the master image group where you can choose from python.
• ... the Image such as base (for python), ml (for python) and conda , pytorch , tensorflow (for python). igh*.
• ... to which of the known nodes you want to send this project (You can choose multiple nodes by clicking on the green +-icon.).
• ... some Data/Parameters information, which will be necessary for the training algorithm in the Data/Parameter - text block.

After clicking on the Create-button the project will be sent to all the selected nodes, so each side can either approve or reject the analysis. How the approval process works from the perspective of a realm admin, see the admin guide.

A analysis is targeted at specific data from different nodes. But data can be quite different depending on combinations of all data. Private medical data in general contains really sensitive information and needs to be handled accordingly. For this, you give a risk estimation, so that each node has an idea about the level of data sensitivity accessed by the analysis. This estimation is categorized into three stages:

• Low risk (green): The analysis won't use any personal related data, such as calculation about all Loinc-Codes of Observation-Resources in a FHIR-Server.
• Mid risk (yellow): The analysis will use only one or hardly any personal information of a patient such as gender or age. One param will not be harmful at all, but should still be mentioned when used during the process.
• High (red): The analysis in general will perform some calculation/analysis on private data where it may be possible that private information will be figured out about a single person. This can occur when the calculations will be performed on a lot of personal information of only a few participants. Overall this should nevertheless not be an issue when the FHIR-Servers provide enough privacy-preserving measurements such as pseudonymization or anonymization.

Approval

Before an analysis can be performed based on your submitted project, it must be approved by the nodes you selected. See the Admin Guide for details on how this process will be handled by the admin.