Examples of AI-assisted MRD analyses. Pipeline exports were analyzed using a conventional FC software. Events were automatically classified and color-coded based on DNN class parameter values, corresponding to CD19⁻ hematogones (gold), early hematogones (green), late hematogones (orange), mature B-cells (violet), and others. True MRD percentages (red boxes) were estimated on downsampled exports using the upsampling factor parameter and simple mathematical operations. (A) Typical MRD population with abnormal bright CD10 expression and loss of CD38, rapidly identified using a single gate on an aberrancy scale vs CD19 dot plot. (B) Less frequent MRD immunophenotype showing bright CD66c expression as the only detectable abnormality, also easily identified using the aberrancy scale parameter. (C) B-ALL MRD with unusual immunophenotypic features (abnormal dim to negative CD45 expression and negativity for CD34), easily identified based on high aberrancy scale values but more accurately delineated using a cluster UMAP gate. (D) Challenging B-ALL MRD population in a patient with a history of anti-CD19 (blinatumomab) and anti-CD20 (rituximab) therapies, showing partial dim expression of CD22 as the only detectable B-lineage antigen (negative for CD10, CD19, CD20, and CD24). This subset was easily identified using a single gate on high aberrancy scale values. SSC-A, side scatter amplitude.