Flexible Group Fairness Metrics for Survival Analysis

TL;DR

This paper investigates how to adapt and apply group fairness metrics to survival analysis, highlighting the effectiveness of discrimination measures and identifying gaps in calibration and scoring rules for bias detection.

Contribution

It introduces flexible group fairness metrics tailored for survival analysis and evaluates their effectiveness across multiple datasets and measures.

Findings

Discrimination measures effectively detect bias in survival models.

Calibration and scoring rules show less clarity in bias measurement.

Empirical validation on 29 datasets demonstrates the utility of proposed metrics.

Abstract

Algorithmic fairness is an increasingly important field concerned with detecting and mitigating biases in machine learning models. There has been a wealth of literature for algorithmic fairness in regression and classification however there has been little exploration of the field for survival analysis. Survival analysis is the prediction task in which one attempts to predict the probability of an event occurring over time. Survival predictions are particularly important in sensitive settings such as when utilising machine learning for diagnosis and prognosis of patients. In this paper we explore how to utilise existing survival metrics to measure bias with group fairness metrics. We explore this in an empirical experiment with 29 survival datasets and 8 measures. We find that measures of discrimination are able to capture bias well whereas there is less clarity with measures of calibration and scoring rules. We suggest further areas for research including prediction-based fairness metrics for distribution predictions.