Hazard Gradient Penalty for Survival Analysis

TL;DR

This paper introduces hazard gradient penalty (HGP), a regularization technique for survival analysis models, especially those based on ODE frameworks, to improve their performance by constraining hazard function gradients.

Contribution

We propose HGP, a novel regularizer for survival models that enhances performance by controlling hazard function gradients, applicable to any model including ODE-based frameworks.

Findings

HGP outperforms existing regularizers on benchmark datasets.

Theoretically, HGP relates to minimizing KL divergence between local density functions.

Applicable to various survival analysis models, including ODE-based ones.

Abstract

Survival analysis appears in various fields such as medicine, economics, engineering, and business. Recent studies showed that the Ordinary Differential Equation (ODE) modeling framework unifies many existing survival models while the framework is flexible and widely applicable. However, naively applying the ODE framework to survival analysis problems may model fiercely changing density function which may worsen the model's performance. Though we can apply L1 or L2 regularizers to the ODE model, their effect on the ODE modeling framework is barely known. In this paper, we propose hazard gradient penalty (HGP) to enhance the performance of a survival analysis model. Our method imposes constraints on local data points by regularizing the gradient of hazard function with respect to the data point. Our method applies to any survival analysis model including the ODE modeling framework and is easy to implement. We theoretically show that our method is related to minimizing the KL divergence between the density function at a data point and that of the neighborhood points. Experimental results on three public benchmarks show that our approach outperforms other regularization methods.