Blogs & News
Aridhia Attends 2025 PHEMS Conference
Last year we blogged about PHEMS’ first annual consortium meeting in London, and the progress we had already made in integrating a federated use case with the Aridhia DRE. This year, Scott Russell, Riccardo Casula, Deepak Sehgal and I attended the 2025 annual meeting in Rotterdam, where Riccardo and I also presented an update on our work to enable the PHEMS federated use cases.
Day One
The highlight of day one was an update on the development of the PHEMS use cases covering both their benchmarking and machine learning requirements. Use case one, benchmarking cardiology outcomes via the PHEMS network is of particular interest to us as it is the most advanced, is being led by Great Ormond Street Hospital, and the MVP networking dashboard is hosted in an Aridhia workspace.
While still under development, it already allows users to compare the outcomes of a variety of different paediatric cardiology procedures at the participating hospitals.
Day one concluded with a discussion on possible future use cases for PHEMS, and how best to take forward the ML requirements in use cases 2 and 3.
Day Two
On day two we presented our update on the development of the PHEMS network, and with assistance of GOSH and HUS were able to provide a live demo of the federated analysis that underpins use case one.
A previous blog details how the PHEMS network operates, but in brief summary GitHub is used to manage the analytical code and initiate federated tasks, and the data controllers deploy their federated nodes in outbound mode to listen for and execute new tasks. At present the results are delivered either to a secure Aridhia workspace or back to Github.
The demo, and accompanying breakdown of the roles involved in the federated data analysis workflow, generated a lively discussion around the organisational requirements for scaling the network which we look forward to continuing over the coming months.
The demo was followed by a workshop where the current data controllers fed back on their experience of deploying the federated node and connecting it to the PHEMS network. SJDs contribution was particularly notable, unlike the other partners they have deployed their node to AWS, and encountered some difficulties. In response we will be making some minor adjustments to the FN to ensure that it can be easily deployed to their environment.
What Next?
As noted above, the next release of the Federated Node will contain changes to ensure ease of deployment for organisations using AWS. Our larger development priority is the integration of a machine learning capability with the PHEMS network, and we will be working on this with our consortium partners for the rest of 2025.