We developed several computational and
machine learning methods to successfully identify behavioral patterns or signatures associated with
different classes of reference drugs, from which to predict the class of novel compounds (Brunner et al., 2012, Alexandrov et al., 2015), and more recently developed
methods to allow us to compare animal models of AD and its progression, and to identify (in silico) novel compounds from our existing database of thousands of novel and reference compounds with the potential to reverse the AD model behavioral profile.