Our aim is to develop a machine learning-based model to predict personalized treatment for relapsed or refractory Acute Leukemia. We will exploit two new and promising Synthetic Lethality approaches by multilayer analysis of omics data. Differential expression analysis of RNA-Seq data will provide primary molecular markers that will be propagated in protein-protein interaction networks, to identify secondary not altered candidates genes of Synthetic Lethality. The identified markers will be reduced by selecting DNA Damage Response (DDR)-related genes for which a drug inhibitor is available. The selected drugs will be screened ex-vivo on primary leukemic cells. Combination of omics and ex-vivo drug screening data will be used as training dataset for the predictive model. A multi-centric investigator-initiated clinical trial will test the performance of the predictive model as well as the practical feasibility of the whole procedure. Our ambition is to precisely match leukemia patients with the most effective targeted therapy, thereby increasing chances of successful treatment.