Fig. 5
From: Multi-omics integration for neuroblastoma clinical endpoint prediction
Autoencoder schematic and survival analysis. a Scheme of the autoencoder architecture giving the best results. b Kaplan-Meier survival curves for the two groups of patients identified clustering TR set autoencoder meta-features significantly associated with survival. c Kaplan-Meier survival curves for the two groups of patients identified clustering TS set autoencoder meta-features significantly associated with survival. The significant p-values suggest that the groups identify two distinct patient populations. Curves for TR/TS patients were calculated separately to highlight that the selection of survival-associated meta-feature in the TR set effectively stratifies also TS set patients