Partly due to our ongoing research efforts as part of SYMPHONY, our researchers in WP06 were contacted to participate in a EU Horizons project. After an initial period of uncertainty – the project was initially denied funding only to be accepted a month later – the PHEMS project (phems.eu) has officially started! The goal of the project is to connect the electronic health record systems of several paediatric hospitals in Europe in order to facilitate collaboration and share sensitive patient data in a privacy respecting manner. To this end, the PHEMS project aims to set up a so-called federated learning environment, where data remains at local institutions and instead prediction models are shared between each treatment centre. This way, the data can be kept private, while models can still learn from all the data. The feasibility and benefits of the approach will be showcased using novel machine learning methods in three clinical use cases. The SYMPHONY team will work on one of these use-cases: we will use recent ML algorithms produced within the SYMPHONY consortium to predict the pharmacokinetic profile of paediatric patients with haemophilia. Something we are very excited about!
A large component of PHEMS will focus on connecting the various healthcare systems of all the partners, something similar we have also been working on in WP07 within the context of different haemophilia treatment centres in the Netherlands. Yet another component of the project is to generate synthetic data based on the data from each of the paediatric hospitals that can be shared with external partners. This, combined with the prospect of sharing data from multiple hospitals, is a great prospect for improving diagnostics and predictive capabilities of models for rare diseases, where we usually do not have access to very large data sets. This of course also aligns very well with the goals of SYMPHONY.
Please keep an eye out for our contributions on the PHEMS website!
By Alexander Janssen (PhD WP06)