Gradient-based Explanations for Deep Learning Survival Models

TL;DR

This paper introduces gradient-based explanation methods for deep survival models, enabling better interpretability of time-dependent features in personalized medicine applications.

Contribution

It extends gradient-based explanation techniques to survival neural networks, analyzing their assumptions and proposing new visualizations and methods like GradSHAP(t).

Findings

Gradient methods effectively capture feature effects and time dependencies.

GradSHAP(t) outperforms SurvSHAP(t) and SurvLIME in speed and accuracy.

Applied to medical data, methods reveal relevant features and temporal patterns.

Abstract

Deep learning survival models often outperform classical methods in time-to-event predictions, particularly in personalized medicine, but their "black box" nature hinders broader adoption. We propose a framework for gradient-based explanation methods tailored to survival neural networks, extending their use beyond regression and classification. We analyze the implications of their theoretical assumptions for time-dependent explanations in the survival setting and propose effective visualizations incorporating the temporal dimension. Experiments on synthetic data show that gradient-based methods capture the magnitude and direction of local and global feature effects, including time dependencies. We introduce GradSHAP(t), a gradient-based counterpart to SurvSHAP(t), which outperforms SurvSHAP(t) and SurvLIME in a computational speed vs. accuracy trade-off. Finally, we apply these methods to medical data with multi-modal inputs, revealing relevant tabular features and visual patterns, as well as their temporal dynamics.