Toward Conditional Distribution Calibration in Survival Prediction

TL;DR

This paper introduces a conformal prediction-based method to improve both marginal and conditional calibration in survival prediction models, emphasizing its importance for individual decision-making and demonstrating its effectiveness across diverse datasets.

Contribution

It presents a novel approach that enhances conditional calibration in survival models using conformal prediction, with theoretical guarantees and extensive empirical validation.

Findings

Improves marginal and conditional calibration without reducing discrimination.

Provides asymptotic theoretical guarantees for calibration.

Demonstrates effectiveness across 15 diverse real-world datasets.

Abstract

Survival prediction often involves estimating the time-to-event distribution from censored datasets. Previous approaches have focused on enhancing discrimination and marginal calibration. In this paper, we highlight the significance of conditional calibration for real-world applications -- especially its role in individual decision-making. We propose a method based on conformal prediction that uses the model's predicted individual survival probability at that instance's observed time. This method effectively improves the model's marginal and conditional calibration, without compromising discrimination. We provide asymptotic theoretical guarantees for both marginal and conditional calibration and test it extensively across 15 diverse real-world datasets, demonstrating the method's practical effectiveness and versatility in various settings.