Machine-learning unterstützte Immunphänotypisierung – Chancen & Limitationen

Program

Immonophenotyping is a central component of hematological diagnostics for classifying and quantifying potentially pathogenic cell populations. Modern flow cytometry devices measure the cell surface expression of up to 50 different markers on tens of thousands of cells, generating large, high-dimensional data sets. Due to the complexity of the data structure, manual analysis of the data is often challenging, subjective and limited to a fraction of the data. An alternative classification based on machine learning benefits from large data volumes and the limitations of machine analysis do not lie in data complexity. As part of the training, I would like to present the basic idea behind machine-learning classification and its application in immunophenotyping using an ongoing research project.