Federated Machine Learning for Systems Medicine

Abstract

Improvements in the wet-lab technologies enable an unprecedented view on how cells and organisms proliferate, adopt, and are influenced by for instance complex disease. Nevertheless, the monolithic view on individual molecular components has only limited explanatory power and underlying mechanisms might remain hidden in the wealth of data. Here, we seek to demonstrate the potential of a system medical approach by integrating the molecular interplay into the analysis and therefore unraveling a mechanistic view on health and diease. Nevertheless, all the methods require high-quality, large amount of data. This puts a huge roadblock on research, even if the data exists since consolidating data in a common cloud fails due to legal and regulatory restrictions. In this talk, the transformation toward federated machine learning as a solution for the data scarcity problem is presented as a potential solution to enable global collaborative precision medicine in a privacy-aware manner.