The Workshop aimed to give an overview of recent methodological developments and their applications in the analysis of complex and imperfect (i.e. realistic) medical data, from epidemiological studies or from clinical trials. 

The emphasis was on the following four themes: overfitting suppression (too many covariates and/or too few samples), data-driven identification of treatment responders, informatively missing covariate or outcome data (e.g. how to decontaminate inference results for informative censoring by competing risks),and federated methods (how to benefit from large datasets without sharing/combining data distributed at different centres). 

We believe the workshop was received positively, and helped creating a common understanding of the potential and limitations of interpretable analysis tools in the rapidly evolving field of personalized medicine.