High-throughput screening approaches, where hundreds of thousands of compounds are evaluated in microamounts for their activity against certain targets, can regularly result in hit rates that are only a fraction of a percent. Here, we take a previously developed machine-learning classification model (with the Signature molecular descriptor) used to identify active compounds against Factor XIa and experimentally verify the virtual screening model predictions. Of 21 predicted compounds tested, seven show activity against Factor XIa, a 33% hit rate. (C) 2014 American Institute of Chemical Engineers