Mr. Cansiz represents the project partner organisation Inria – one of the largest national research institutes in France, with numerous research teams working on a wide range of topics in digital science and technology. Sergen is currently involved in the development of the federated learning software Fed-BioMed.

𝐐: 𝐈𝐧 𝐲𝐨𝐮𝐫 𝐯𝐢𝐞𝐰, 𝐰𝐡𝐚𝐭 𝐰𝐢𝐥𝐥 𝐛𝐞 𝐭𝐡𝐞 𝐦𝐨𝐬𝐭 𝐬𝐢𝐠𝐧𝐢𝐟𝐢𝐜𝐚𝐧𝐭 𝐢𝐦𝐩𝐚𝐜𝐭 𝐨𝐟 𝐭𝐡𝐞 𝐃𝐓𝐑𝐈𝐏4𝐇 𝐢𝐧𝐢𝐭𝐢𝐚𝐭𝐢𝐯𝐞 𝐨𝐧 𝐭𝐡𝐞 𝐟𝐮𝐭𝐮𝐫𝐞 𝐨𝐟 𝐡𝐞𝐚𝐥𝐭𝐡 𝐫𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐢𝐧 𝐄𝐮𝐫𝐨𝐩𝐞?
A: DTRIP4H is a project that brings together various technologies to enable research infrastructures, small and medium-sized enterprises, hospitals, and institutions to collaborate effectively in creating digital twins for use in various healthcare domains. I believe the most impactful outcome of the DTRIP4H project will be the development of a secure, scalable, interoperable, and AI-enabled decentralized digital twin environment. This environment will allow various types of users, such as engineers, developers, students, researchers, and clinicians, to collaborate effectively in creating and using digital twins.

𝐐: 𝐖𝐡𝐚𝐭 𝐢𝐬 𝐈𝐍𝐑𝐈𝐀’𝐬 𝐦𝐚𝐢𝐧 𝐫𝐨𝐥𝐞 𝐢𝐧 𝐭𝐡𝐞 𝐃𝐓𝐑𝐈𝐏4𝐇 𝐩𝐫𝐨𝐣𝐞𝐜𝐭?
A: Inria is one of the technical leaders of the project. Our main role is to provide software and expertise for the federated learning infrastructure for health care in DTRIP4H project.
The software Inria brings to the project is called Fed-BioMed. It is developed by one of Inria’s research teams, Epione, whose work focuses on a range of e-medicine applications in healthcare, and Fed-BioMed initially is designed to support these use cases by enabling privacy-preserving, decentralized collaborative machine learning.
In addition, Inria is leading Work Package 6, which focuses on designing a reference architecture for a decentralized digital twin environment. In this context, we are contributing our expertise in decentralized collaborative machine learning applications in healthcare.

𝐐: 𝐂𝐨𝐮𝐥𝐝 𝐲𝐨𝐮 𝐩𝐥𝐞𝐚𝐬𝐞 𝐞𝐱𝐩𝐥𝐚𝐢𝐧, 𝐢𝐧 𝐬𝐢𝐦𝐩𝐥𝐞 𝐭𝐞𝐫𝐦𝐬, 𝐰𝐡𝐚𝐭 𝐅𝐞𝐝𝐞𝐫𝐚𝐭𝐞𝐝 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐢𝐬 𝐚𝐧𝐝 𝐡𝐨𝐰 𝐢𝐭 𝐜𝐚𝐧 𝐛𝐞 𝐚𝐩𝐩𝐥𝐢𝐞𝐝 𝐢𝐧 𝐭𝐡𝐞 𝐡𝐞𝐚𝐥𝐭𝐡𝐜𝐚𝐫𝐞 𝐬𝐞𝐜𝐭𝐨𝐫?
A: Classical machine learning methods focus on centralizing data to perform training of AI models. However, federated learning aims to enable the training without centralizing the data. The motivation behind this approach is to preserve the privacy of data owners. This is especially important in healthcare, where data is highly sensitive and private. In federated learning, the data stays local — typically within hospitals or research infrastructures in the healthcare context. Instead of sharing the data, AI models are trained locally at each site. Only the model updates are shared and then aggregated to build a final global model. This approach enables collaboration between different infrastructures and allows the use of more data to develop more accurate AI models for healthcare.

𝐐: 𝐖𝐡𝐚𝐭 𝐚𝐬𝐩𝐞𝐜𝐭𝐬 𝐨𝐟 𝐈𝐍𝐑𝐈𝐀’𝐬 𝐜𝐨𝐧𝐭𝐫𝐢𝐛𝐮𝐭𝐢𝐨𝐧 𝐭𝐨 𝐭𝐡𝐢𝐬 𝐩𝐫𝐨𝐣𝐞𝐜𝐭 𝐝𝐨 𝐲𝐨𝐮 𝐟𝐢𝐧𝐝 𝐦𝐨𝐬𝐭 𝐬𝐢𝐠𝐧𝐢𝐟𝐢𝐜𝐚𝐧𝐭 𝐨𝐫 𝐢𝐧𝐬𝐩𝐢𝐫𝐢𝐧𝐠?
A: First of all, as Inria and Fed-BioMed team, we are motivated to see Fed-BioMed being used for federated learning in this project. As we participate in new projects, we gain experience across different areas of healthcare. This gives us the opportunity to enhance existing functionalities, address more use cases, and improve the overall user experience.
This project is particularly interesting because it brings together a variety of use cases. Integrating Fed-BioMed into a decentralized digital twin environment to apply federated learning across these diverse scenarios is something we find especially exciting.

𝐐: 𝐖𝐡𝐚𝐭 𝐝𝐨 𝐲𝐨𝐮 𝐭𝐡𝐢𝐧𝐤 𝐚𝐫𝐞 𝐭𝐡𝐞 𝐤𝐞𝐲 𝐚𝐝𝐯𝐚𝐧𝐭𝐚𝐠𝐞𝐬 𝐨𝐟 𝐰𝐨𝐫𝐤𝐢𝐧𝐠 𝐰𝐢𝐭𝐡𝐢𝐧 𝐭𝐡𝐢𝐬 𝐦𝐮𝐥𝐭𝐢𝐝𝐢𝐬𝐜𝐢𝐩𝐥𝐢𝐧𝐚𝐫𝐲 𝐄𝐮𝐫𝐨𝐩𝐞𝐚𝐧 𝐜𝐨𝐧𝐬𝐨𝐫𝐭𝐢𝐮𝐦?
A: I think it's especially important for us to work within a multidisciplinary consortium, because the software we develop and maintain at Inria — Fed-BioMed — is designed to support various fields within healthcare. Collaborating with experts from different domains allows us to explore and experience diverse perspectives, which broadens our vision and helps us better understand a wide range of healthcare applications.

𝘞𝘦 𝘴𝘪𝘯𝘤𝘦𝘳𝘦𝘭𝘺 𝘵𝘩𝘢𝘯𝘬 𝘔𝘳. 𝘊𝘢𝘯𝘴𝘪𝘻 𝘧𝘰𝘳 𝘴𝘩𝘢𝘳𝘪𝘯𝘨 𝘩𝘪𝘴 𝘦𝘹𝘱𝘦𝘳𝘵𝘪𝘴𝘦 𝘰𝘯 𝘧𝘦𝘥𝘦𝘳𝘢𝘵𝘦𝘥 𝘭𝘦𝘢𝘳𝘯𝘪𝘯𝘨 𝘢𝘯𝘥 𝘥𝘪𝘨𝘪𝘵𝘢𝘭 𝘵𝘸𝘪𝘯 𝘵𝘦𝘤𝘩𝘯𝘰𝘭𝘰𝘨𝘪𝘦𝘴, 𝘢𝘯𝘥 𝘸𝘦 𝘸𝘪𝘴𝘩 𝘩𝘪𝘮 𝘦𝘷𝘦𝘳𝘺 𝘴𝘶𝘤𝘤𝘦𝘴𝘴 𝘪𝘯 𝘵𝘩𝘦 𝘤𝘰𝘮𝘪𝘯𝘨 𝘺𝘦𝘢𝘳 𝘢𝘴 𝘵𝘩𝘦 𝘱𝘳𝘰𝘫𝘦𝘤𝘵 𝘢𝘥𝘷𝘢𝘯𝘤𝘦𝘴 𝘵𝘰𝘸𝘢𝘳𝘥 𝘪𝘵𝘴 𝘯𝘦𝘹𝘵 𝘮𝘪𝘭𝘦𝘴𝘵𝘰𝘯𝘦𝘴!

𝐖𝐞 𝐢𝐧𝐯𝐢𝐭𝐞 𝐲𝐨𝐮 𝐭𝐨 𝐰𝐚𝐭𝐜𝐡 𝐭𝐡𝐞 𝐕𝐈𝐃𝐄𝐎 𝐞𝐝𝐢𝐭𝐢𝐨𝐧 𝐨𝐟 𝐭𝐡𝐢𝐬 𝐢𝐦𝐩𝐚𝐜𝐭𝐟𝐮𝐥 𝐢𝐧𝐭𝐞𝐫𝐯𝐢𝐞𝐰 𝐨𝐧 𝐨𝐮𝐫 𝐘𝐨𝐮𝐓𝐮𝐛𝐞 𝐜𝐡𝐚𝐧𝐧𝐞𝐥 --> https://www.youtube.com/watch?v=SA4z__HtDVw