Despite advancements in the clinical management of paediatric acute myeloid leukaemia (AML), approximately 30% of patients still face relapse. The assessment of measurable residual disease (MRD) has emerged as an essential tool for patient management and risk stratification. However, current MRD methods, including RT-PCR and multiparameter flow cytometry (MFC), face significant limitations. These include the absence of abnormal immunophenotypes in some patients, changes in immunophenotypes over time, and high inter- and intrapatient heterogeneity, leading to relapses in patients initially deemed MRD-negative. Moreover, the lack of universal genetic MRD targets in paediatric AML further complicates effective monitoring. To address these challenges, we aim to develop a computational framework using single-cell omics (epigenome, proteome, and transcriptome) and machine learning for precision characterization of leukaemia stem cells (LSCs) within subgroups of paediatric AML diagnostic categories. Our approach will identify LSC-specific genetic, epigenetic, and surface antigen markers that are tailored to the unique biology of each subgroup within the diagnostic categories of AML.