A Federated Cox Model with Non-Proportional Hazards

TL;DR

This paper introduces a federated Cox model that handles distributed healthcare data and relaxes the proportional hazards assumption, enabling more flexible survival analysis without extensive prior specification.

Contribution

It proposes a novel federated Cox model that accommodates non-proportional hazards and time-varying effects without explicit pre-specification, suitable for secure healthcare data environments.

Findings

Federated model performs comparably to standard Cox models on clinical datasets.

The model relaxes the proportional hazards assumption, allowing for time-varying effects.

Reduces organizational costs by not requiring explicit effect specifications.

Abstract

Recent research has shown the potential for neural networks to improve upon classical survival models such as the Cox model, which is widely used in clinical practice. Neural networks, however, typically rely on data that are centrally available, whereas healthcare data are frequently held in secure silos. We present a federated Cox model that accommodates this data setting and also relaxes the proportional hazards assumption, allowing time-varying covariate effects. In this latter respect, our model does not require explicit specification of the time-varying effects, reducing upfront organisational costs compared to previous works. We experiment with publicly available clinical datasets and demonstrate that the federated model is able to perform as well as a standard model.