A Generalizable AI Framework for Predicting Clinically Actionable Molecular Alterations and Prognosis in Acute Myeloid Leukemia: Development and Randomized Trial Validation

This dataset contains de-identified data used to validate the LEAP (Leukemia End-to-End Analysis Platform) framework. The dataset is organized as follows: • Mutation file — Includes the cross-validation splits and the pre-extracted LEAP embeddings from all LEAP Experts for all experiments related to mutation prediction (PML-RARA fusion, NPM1, FLT3, ASXL1, CEBPA, TP53, and complex karyotype). • Survival file — Contains the clinical metadata used for the survival analyses. • Clinical Trial file — Contains the de-identified responses from all pathologists who participated in our clinical trial (ClinicalTrials.gov, NCT07203885)