The aim of this MISE financed project is to create a prototype of a federated learning infrastructure for clinical and omic data. While machine learning is increasingly becoming a key asset for rationalizing clinical and omic data, the practical application of these methodologies bring to the stage the patients privacy preservation issue. Federated learning represents a technique to both take advantage of several hospitals data and at the same granting that data never leaves the hospitals themselves, hence rendering the approach privacy-preserving by design. In this project we will deal with omic data including: metabolomics, genomics, lipidomics, imageomics and clinical data. This project has several industrial and clinical partners which will bring their own expertise and data to validate the prototype.