Dual Mixture-of-Experts Framework for Discrete-Time Survival Analysis

TL;DR

This paper introduces a dual mixture-of-experts framework for discrete-time survival analysis that improves predictive accuracy by modeling patient heterogeneity and temporal dynamics, demonstrating enhanced performance on breast cancer datasets.

Contribution

It presents a novel dual MoE architecture combining subgroup-aware representation and temporal hazard modeling, adaptable to existing deep survival analysis pipelines.

Findings

Improves time-dependent C-index by up to 0.04 on breast cancer datasets.

Enhances performance when integrated into the ConSurv framework.

Flexible design that can be incorporated into existing deep learning survival models.

Abstract

Survival analysis is a task to model the time until an event of interest occurs, widely used in clinical and biomedical research. A key challenge is to model patient heterogeneity while also adapting risk predictions to both individual characteristics and temporal dynamics. We propose a dual mixture-of-experts (MoE) framework for discrete-time survival analysis. Our approach combines a feature-encoder MoE for subgroup-aware representation learning with a hazard MoE that leverages patient features and time embeddings to capture temporal dynamics. This dual-MoE design flexibly integrates with existing deep learning based survival pipelines. On METABRIC and GBSG breast cancer datasets, our method consistently improves performance, boosting the time-dependent C-index up to 0.04 on the test sets, and yields further gains when incorporated into the Consurv framework.