Optimal Cox regression under federated differential privacy: coefficients and cumulative hazards

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hung, Elly K. H., Yu, Yi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914444034965504
author Hung, Elly K. H.
Yu, Yi
author_facet Hung, Elly K. H.
Yu, Yi
contents We study two foundational problems in distributed survival analysis under federated differential privacy (FDP): estimation of the Cox regression coefficients and of the cumulative baseline hazard functions, allowing for heterogeneous per-sever sample sizes and privacy budgets. To quantify the fundamental cost of privacy, we derive minimax lower bounds together with upper bounds that match up to poly-logarithmic factors for the regression coefficients, thereby revealing server-level phase transitions between private and non-private regimes. We also consider a relaxed differential privacy framework with partially public information. Our analysis shows that the role of public covariates depends strongly on the privacy model. For cumulative hazard estimation, we propose a private tree-based version of the Breslow estimator for nonparametric integral estimation under FDP. As a by-product, this leads to a private survival function estimator that attains a nearly minimax optimal rate. Numerical experiments, including a real-data application, support the theoretical findings. The proposed methods are implemented in an accompanying R package FDPCox.
format Preprint
id arxiv_https___arxiv_org_abs_2508_19640
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Optimal Cox regression under federated differential privacy: coefficients and cumulative hazards
Hung, Elly K. H.
Yu, Yi
Statistics Theory
Methodology
We study two foundational problems in distributed survival analysis under federated differential privacy (FDP): estimation of the Cox regression coefficients and of the cumulative baseline hazard functions, allowing for heterogeneous per-sever sample sizes and privacy budgets. To quantify the fundamental cost of privacy, we derive minimax lower bounds together with upper bounds that match up to poly-logarithmic factors for the regression coefficients, thereby revealing server-level phase transitions between private and non-private regimes. We also consider a relaxed differential privacy framework with partially public information. Our analysis shows that the role of public covariates depends strongly on the privacy model. For cumulative hazard estimation, we propose a private tree-based version of the Breslow estimator for nonparametric integral estimation under FDP. As a by-product, this leads to a private survival function estimator that attains a nearly minimax optimal rate. Numerical experiments, including a real-data application, support the theoretical findings. The proposed methods are implemented in an accompanying R package FDPCox.
title Optimal Cox regression under federated differential privacy: coefficients and cumulative hazards
topic Statistics Theory
Methodology
url https://arxiv.org/abs/2508.19640