Skip to main content
. 2021 Dec 8;7:60. doi: 10.1038/s41537-021-00191-y

Fig. 5. Architecture of autoencoder.

Fig. 5

The autoencoder takes as input the bio-social features. The output layer is used to reconstruct the input. The hidden layer of the autoencoder is used for dimensionality reduction. We use the hidden layer as input to a random forest model to predict mortality. The hidden layer is composed of 10 neurons.