From Data to Diagnosis: How Federated Learning Protects Patient Privacy While shaping the Future of Healthcare

Machine learning is revolutionizing healthcare by creating new possibilities for clinical research and patient care. However, its effectiveness relies on access to large, diverse datasets, which often necessitates sharing sensitive patient information across institutions. This raises significant concerns about privacy and security, challenging the full realization of machine learning’s potential in healthcare. 

The European Infrastructure for Translational Medicine (EATRIS) and CERN are jointly organizing this webinar to explore how federated learning can address these challenges by allowing the training of machine learning models across multiple institutions without sharing patient data. Through federated learning, algorithms are trained locally on decentralized data, with only model updates being shared and aggregated. This approach preserves patient privacy while leveraging extensive datasets that are crucial for developing accurate and effective predictive models. 

EATRIS will emphasize the importance of multidisciplinary collaboration in driving digitalization and artificial intelligence in healthcare, bringing in the example of TRUSTroke, a European project aimed at optimizing stroke treatment using AI and federated learning. In particular, Carlos Molina, Director of Vall d’Hebron Stroke Center at Vall d’Hebron Research Hospital in Barcelona, Spain, and TRUSTroke coordinator will describe specific challenges in managing stroke patients and assessing disease progression using clinical data. Luigi Serio, Principal Scientist at CERN will illustrate how a federated learning platform developed and operated at CERN can provide robust machine learning algorithms trained across several data silos without exchanging any personal information. Additionally, the webinar will provide insights from Siemens Healthineers, discussing current gaps in the patient pathway and how the industry can help bridge these gaps to improve patient outcomes. 

The event will conclude with a round table discussion and Q&A session, providing participants with the opportunity to engage with experts on the ethical, technical, and practical considerations involved in the implementation of federated learning in healthcare. 

Furthermore, the webinar will include an interactive questionnaire throughout, allowing us to gauge the audience’s opinions on this new technology.

Watch the recording of the webinar below:

Preliminary agenda: 

• Welcome and introduction – Anton Ussi, Operations & Finance Director at EATRIS 
• Multidisciplinary collaboration as enablers of digitalisation and AI – Sara Zullino, Scientific & SME Outreach Manager at EATRIS 
• The challenge in stroke treatment optimisation – Carlos Molina, Director of the Stroke Unit and Brain Hemodynamics at Vall d’Hebron Research Hospital and TRUSTroke coordinator 
• CAFEIN: CERN’s federated learning solution for the privacy-preserving training of AI algorithms – Luigi Serio, Principal Scientist at CERN 
• Industry perspective: current efforts and bridging the gap: how artificial intelligence can be used for better stroke patient outcomes – Hanno Herrmann, Siemens Healthineers
• Round table discussion and Q&A (Moderator: Anton Ussi – EATRIS)