AI pipeline for FC data. (A) Pipeline scheme showing intake of unaltered FC (fcs) files produced by the analyzer, main processing steps, and AI-enhanced downsampled exports with added AI parameters. The pipeline was trained with 31 negative controls samples, which were used as a control template to measure deviation from normal, and also gated by expert analysts to train a DNN event classifier. (B) A cluster-based aberrancy scale was used to quantify deviation from normal, based on merging sample and control events, high-resolution clustering, and inverse log-scaling the fraction of sample events per cluster. FCS, fcs files.