Interpretable Prediction and Feature Selection for Survival Analysis

TL;DR

This paper introduces DyS, an interpretable and accurate survival analysis model that combines feature selection with prediction, especially suited for large healthcare datasets, balancing discrimination and interpretability.

Contribution

DyS is a novel generalized additive model that integrates feature selection and interpretable prediction for survival analysis, addressing the need for transparency in healthcare applications.

Findings

DyS achieves competitive discrimination performance compared to state-of-the-art models.

DyS uses fewer features, enhancing interpretability.

Empirical results demonstrate DyS's effectiveness on large healthcare datasets.

Abstract

Survival analysis is widely used as a technique to model time-to-event data when some data is censored, particularly in healthcare for predicting future patient risk. In such settings, survival models must be both accurate and interpretable so that users (such as doctors) can trust the model and understand model predictions. While most literature focuses on discrimination, interpretability is equally as important. A successful interpretable model should be able to describe how changing each feature impacts the outcome, and should only use a small number of features. In this paper, we present DyS (pronounced ``dice''), a new survival analysis model that achieves both strong discrimination and interpretability. DyS is a feature-sparse Generalized Additive Model, combining feature selection and interpretable prediction into one model. While DyS works well for all survival analysis problems, it is particularly useful for large (in $n$ and $p$) survival datasets such as those commonly found in observational healthcare studies. Empirical studies show that DyS competes with other state-of-the-art machine learning models for survival analysis, while being highly interpretable.