PhD defence Varsha Gouthamchand

Keywords: F.A.I.R. Data, Semantic Web, Knowledge Graphs, Federated Learning

"From Heterogeneity to Harmony: F.A.I.R. Integration for Federated Big Data Analyses in Oncology"

This thesis explores how large amounts of medical data can be safely shared and analysed to improve cancer care. In modern oncology, hospitals collect vast amounts of information, spread across imaging systems, test results, and EHR systems. However, differences in data formats and strict privacy rules often make it difficult to combine this knowledge. This research addresses these challenges by applying the F.A.I.R. (Findable, Accessible, Interoperable, Reusable) principles and privacy-enhancing methods. The developed tools allow hospitals to work together without exchanging sensitive patient data, using a method called Federated Learning. By applying these techniques to Head & Neck and Lung cancer data, this work shows how reliable prediction models can be built across multiple institutions while protecting patient privacy. With this, the thesis provides practical insights for future applications of Artificial Intelligence (AI) in healthcare. 

Click here for the live stream.