From September 12 to 14, 2024, an important international meeting in the Healthcare field is held in Belgrade: the Caring for Rare conference: the event organized by NORBS to foster collaboration between experts and organizations from Central and Eastern Europe.
Matteo Bregonzio, CTO and head of the Research and Development team at Datrix, as a keynote speaker in the session “Empowering Rare Diseases through Data” discussed the positive impact of AI-based technologies in the field of rare diseases: not only increasingly early diagnoses and new personalized therapeutic treatments, but also access to a quantity of insights and evidence that would be impossible to reach with traditional methods.
The focus of the intervention was the BETTER project (a project with 10 million euros in EU funding, of which Datrix is the international coordinator) which aims to provide doctors in European hospitals with a consultation and decision support interface based on genomic analysis via AI. Thanks to Federated Learning technology, of which Datrix is a technological enabler, it is possible to aggregate models trained locally without the need to centralize sensitive data in a single database, thus in full compliance with privacy and the AI Act.
You can find an interesting in-depth study in this interview on our platform. On this occasion, we had the pleasure of a chat with Matteo regarding his speech and one of Datrix’s flagship projects.
The positive impact of AI-based technologies on the social fabric
Datrix participated in Caring for Rare. What contribution did you bring to the event as head of R&D at Datrix?
For Datrix, this is an important opportunity to show how the use of AI-based technologies can effectively improve the diagnosis and treatment of rare diseases.
The conference focused on the importance of leveraging data in promoting understanding of molecular pathways, early diagnosis, and the development of innovative treatments. The increasing use of data-driven medicine has made it essential to create secure and privacy-respecting infrastructures for sharing health data at the European level, one of the objectives of the European BETTER project of which Datrix is the leader.
Can you explain what the BETTER project is about and what its goals are?
BETTER is a healthcare project that aims to improve early diagnosis and personalized therapeutic treatments for rare diseases.
R&D activities focus on the collection and analysis of data regarding rare pediatric diseases: over the next three years, the project will work on collecting and analyzing data in three areas: pediatric intellectual disability with particular attention to metabolic diseases from neonatal screening, hereditary retinal dystrophies (including Best macular dystrophy, X-linked retinitis pigmentosa, Stargardt disease), and autism spectrum disorders.
The project aims to revolutionize the care and early diagnosis of these pathologies, using AI to offer more effective and targeted therapies, while respecting global privacy regulations regarding health information.
Precision medicine takes into account individual characteristics and variability at the genetic, metabolic, and environmental levels of the patient; offering a personalized, effective therapeutic path and minimizing side effects. Thanks to the possibility of pooling data from multiple hospitals and studying them through AI, precision medicine will allow the national health service to be more efficient and offer frontier therapies to patients. This is the ambition of the Better project that we will work on for the coming years.
Federated Learning: sensitive data privacy at the center of Datrix R&D
Speaking of data privacy, one of the central themes of your speech was innovation in the processing of sensitive data. Can you tell us more?
Exactly, one of the central themes of the intervention concerned the Personal Health Train (PHT) model and the way it guarantees data confidentiality while facilitating access for research.
Better stands out for its approach of “Federated Learning” and “Distributed AI,” which allows it to comply with stringent global privacy and GDPR regulations, while ensuring secure access to data for research.
Federated Learning allows the training of artificial intelligence models in a distributed manner, without the need to transfer or share sensitive data. In this way, patient data remains within individual healthcare facilities, while local models are trained and subsequently aggregated to create a global model with superior performance.
However, a peculiarity of this technology is that it is not relegated only to the healthcare field but can, instead, be exploited in other contexts where data sensitivity is crucial, such as in the banking, financial, and other regulated industries, where privacy and security are fundamental.
For further information:
