EHDEN Setup of ARACHNE Central and ARACHNE Data Node

[PRijnbeek]

Let me know if I am right:

1) EHDEN will have one central ARACHNE that all data sources we will include will get access to. 2) No data partners have to install ARACHNE central unless they want to build their own sub-network. This could for example be the EU-ADR network that contains 6 data sources for which they want to do their own network studies not public to all others. Another example is IQVIA that will have their own network. (I still need to understand better how the request to another network will work) 3) A Data Partner does not have to install a Data Node on the public internet to be part of the network because he can create a Virtual Database. 4) If they install a data node they can control if they share Achilles results that those that have access to the data node can see (I control access to my data node through Arachne Central). That Data Node then still does not have to be connected to the database because we use the json Achilles version. 5) Study Requests are send as notifications in Arachne Central and they are not (yet) send as emails to the Data Partners. 6) For databases that are completely disconnected from the internet the following solutions would work:

• I create a virtual data source under my account for each one I have access to. When I get a request, i download the code, bring it into my local network, run the code (use a tool,R, or an internal Data Node setup), check the results, bring the results to a machine that has internet access, upload the results back to Arachne Central. Advantage: no Data Node install needed. Disadvantage: I cannot make Achilles results available in Arachne (not sure yet this is what we want or this should be in the database catalogue.
• I setup a publicly accessible Data Node in which I could also link a public Atlas if I like to share cohorts / study definitions or I like to make Achilles results available to users. I follow the same procedure as in a. DIsadvantage: I need to host and maintain Data Node. Advantage: I can share more (Achilles, Studies etc). 7) For the Open Source version, it is not possible to the (semi-)automatic execution (only possible if the Data Node is publicly accessible, and the database is in some secure way connected to the Data Node). In the federated network I do not think this is a big issue because there will be very few data sources that will allow this (for now). However, it may be possible for some Hospitals that are already doing this using other solutions (Mostly simple counts).

.

[PRijnbeek]

@gklebanov can you answer these questions (I assigned Konstantin earlier).

[PRijnbeek]

@gklebanov @konstjar can you please follow-up on these? 2 year old but still valid questions..