Treatment Response Prediction in Acute Myeloid Leukemia Patients

Abstract

Predicting acute myeloid leukemia (AML) patient treatment response has the potential to impact clinical decisions. A prediction given at the time of diagnosis for treatment response can assist physicians in making effective treatment decisions and improving patient prognosis. This project aimed to develop methods that leverage domain knowledge in AML to identify biomarkers and build predictive models from biological datasets. Specifically, we applied our methods to messenger RNA (mRNA) expression and gene mutation data extracted from bone marrow or peripheral blood samples taken at patients' time of diagnosis. Identified biomarkers are used as feature sets to train a prediction model of patients' treatment responsiveness. This prediction will aid physicians in optimizing treatment decisions for patients on an individual basis.