Flow Cytometry (FC) is a widespread technique enabling the study of targeted cell populations in health and disease. FC is however limited in the number of assayed proteins, preventing its application for the generation of “cellular atlases” of complex tissues. In the past, I have shown that combining FC with non-linear multivariate regression Machine Learning (ML) techniques allowed to profile hundreds of proteins (>x10 compared to conventional FC) across millions of single cells, demonstrating that FC can lead to practically fruitful ML applications. This project contains three objectives aiming at applying and further developing this technique : i. Apply this technique to human inflamed or non-inflamed liver samples to obtain an exhaustive cellular profiling ii. Develop supervised classification algorithms for the automated phenotyping of FC datasets leveraging a pre-annotated reference dataset iii. Extend this technique to imaging flow cytometry datasets by using generative models to predict the cellular localisation and spatialized expression levels of hundreds of proteins This project will therefore enable both the generation of new functional hypotheses on proteins expressed by cells subsets of the human liver in the context of chronic inflammation, as well as the development of new ML applications in the context of FC and imaging FC.