Doubly Robust Conformalized Survival Analysis with Right-Censored Data

TL;DR

This paper introduces a conformal inference method for survival analysis with right-censored data, utilizing machine learning for imputation and weighted conformal calibration, supported by asymptotic double robustness, improving predictive inference especially when models are imperfect.

Contribution

It extends conformal inference to right-censored survival data using imputation and weighted calibration, with theoretical double robustness and practical robustness in challenging scenarios.

Findings

Provides more informative predictive bounds for survival times.

Demonstrates robustness in scenarios with model misspecification.

Shows effectiveness on both simulated and real datasets.

Abstract

We present a conformal inference method for constructing lower prediction bounds for survival times from right-censored data, extending recent approaches designed for more restrictive type-I censoring scenarios. The proposed method imputes unobserved censoring times using a machine learning model, and then analyzes the imputed data using a survival model calibrated via weighted conformal inference. This approach is theoretically supported by an asymptotic double robustness property. Empirical studies on simulated and real data demonstrate that our method leads to relatively informative predictive inferences and is especially robust in challenging settings where the survival model may be inaccurate.