Integrating Patient Level Multi-omics Data Using Deep Learning Models

Explore cutting-edge approaches to integrating patient-level multi-omics data using deep learning models in this 50-minute workshop talk by Simon Rasmussen from the NNF Center for Protein Research at the University of Copenhagen and the NNF Center for Genomic Mechanisms of Disease at the Broad Institute. Delve into supervised learning techniques for predicting patient outcomes and discover the power of EIR for large-scale genomics data analysis. Examine methods for integrating genomics and biomarkers, and learn about unsupervised deep learning approaches, including autoencoders, for data integration in a T2D cohort. Gain valuable insights into latent representation and future perspectives in multi-modal data integration, essential for advancing personalized medicine and genomic research.

Intro
Topic: Multi-modal data integration
Supervised: Predict patient outcomes
Deep Learning for integration
EIR: Supervised leaming from large scale genomics data
Integrating genomics and biomarkers
Using EIR to model the biomarkers
Unsupervised DL for data integration
T2D cohort with multi-modal data
Unsupervised deep learning: Autoencoders
Latent representation
Perspectives