Variational Deep Survival Machines: Survival Regression with Censored Outcomes

TL;DR

A novel method to predict the survival time by better clustering the survival data and combine primitive distributions is presented, and two variants of variational auto-encoder, discrete and continuous, are proposed to generate the latent variables for clustering input covariates.

Abstract

Survival regression aims to predict the time when an event of interest will take place, typically a death or a failure. A fully parametric method [18] is proposed to estimate the survival function as a mixture of individual parametric distributions in the presence of censoring. In this paper, We present a novel method to predict the survival time by better clustering the survival data and combine primitive distributions. We propose two variants of variational auto-encoder (VAE), discrete and continuous, to generate the latent variables for clustering input covariates. The model is trained end to end by jointly optimizing the VAE loss and regression loss. Thorough experiments on dataset SUPPORT and FLCHAIN show that our method can effectively improve the clustering result and reach competitive scores with previous methods. We demonstrate the superior result of our model prediction in the long-term. Our code is available at https://github.com/qinzzz/auton-survival-785.

Figures (5)

• Figure 1: Overview of our proposed VDSM.
• Figure 2: The structure of categorical VAE.
• Figure 3: The structure of generative clustering VAE.
• Figure 4: Clustering Result on SUPPORT dataset
• Figure 5: Clustering Result on FLCHAIN dataset