Abstract: Generative Adversarial Networks (GANs) represent a promising class of generative networks that combine neural networks with game theory. From generating realistic images to assisting music creation, GANs are transforming many fields of arts and sciences. In this session, we present a method based on the medGAN architecture to evaluate drug-induced laboratory test changes across time. We demonstrated that it is feasible to generate artificial laboratory test time series that account for drug effects.
Learning Objective 1: At the end of the session, the learner will be able to define generative adversarial networks and will be more familiar with the potential application of these algorithms in biomedical informatics to generate synthetic datasets trained from real data.
Alexandre Yahi (Presenter)
Rami Vanguri, Columbia University
Noemie Elhadad, Columbia University
Nicholas Tatonetti, Columbia University