Extended Data Fig. 3: DLVPM path model used to combine multiomic and imaging data.
From: Integrating multimodal cancer data using deep latent variable path modelling
The figure illustrates the neural network models used as measurement models for each data type in the present investigation. The figure shows the neural network used to process histology data. There are three inputs to the network. These inputs take imaging data that has been tiled, passed through a pre-trained DLVPM-Twins model, and then mean averaged over the network output for each tile. Each of the omics data modalities was processed with a residual network that combines an attentional mechanism with the raw input via a skip connection.
