Fig. 5From: Deep learning based feature-level integration of multi-omics data for breast cancer patients survival analysisMulti-omics data integration with cross-modality autoencoder (CrossAE). For hidden features of each data modality, they are used to reconstruct input features of both the original modality and other modalities. The hidden features of various modalities are element-wise averaged before feeding into the task-specific networkBack to article page