Abstract 2791: Deep learning-based survival prediction of post-operative GBM patients via multimodal radiomic phenotypes drawn from Kaniadakis vector embedding in latent space.