DAGSurv: Directed Acyclic Graph Based Survival Analysis Using Deep Neural Networks

TL;DR

DAGSurv introduces a novel deep neural network approach that leverages causal structures via DAGs for improved survival analysis, outperforming traditional methods on synthetic and real datasets.

Contribution

The paper proposes DAGSurv, a variational autoencoder-based method that incorporates causal DAG knowledge into survival prediction, enhancing accuracy over existing models.

Findings

DAGSurv outperforms Cox, DeepSurv, and Deephit on multiple datasets.

Incorporating causal structure improves survival prediction accuracy.

Method effective in both low and high-dimensional data contexts.

Abstract

Causal structures for observational survival data provide crucial information regarding the relationships between covariates and time-to-event. We derive motivation from the information theoretic source coding argument, and show that incorporating the knowledge of the directed acyclic graph (DAG) can be beneficial if suitable source encoders are employed. As a possible source encoder in this context, we derive a variational inference based conditional variational autoencoder for causal structured survival prediction, which we refer to as DAGSurv. We illustrate the performance of DAGSurv on low and high-dimensional synthetic datasets, and real-world datasets such as METABRIC and GBSG. We demonstrate that the proposed method outperforms other survival analysis baselines such as Cox Proportional Hazards, DeepSurv and Deephit, which are oblivious to the underlying causal relationship between data entities.