Relapse prediction in leukemia

The goal of the study was to predict wether flow cytometry data at diagnosis could improve risk stratification in children with B-cell Acute Lymphoblastic Leukemia. Conclusion: Flow cytometry data, as characterized by the profile of intensity of expression, is unable to predicte relapse. Alternative characterizations are required to keep exploring this research direction. Notes: In this study we characterize the marker expression by summarizing each marker's intensity of expression with basic statistical moments: Mean, median, standard deviation, skewness and kurtosis. We feed this information into a classifier to obtain an estimate of its predictive capacity. Code and patient annotations available at https://github.com/Almr95/Relapse-Prediction Standard quality control for FACsCanto II. Manual and computational preprocessing was performed (doublets, debris, margin events, acquisition errors, batch effects)