A Filtered Mixture-of-Generators for Fully Synthetic Survival Training

2026-06-30

AI Summary

This study addresses the challenge of scarce, costly, and privacy-constrained real-world training data in clinical survival analysis by reframing synthetic data generation as a sample selection problem. The authors propose a two-stage optimization framework: first, a diverse candidate pool is constructed using four heterogeneous tabular generative models; then, high-quality samples are selected via an ensemble scoring mechanism comprising seven models trained on real data, guided by proper scoring rules and a stratified balancing strategy. Evaluated across 16 public datasets, the approach yields an average C-index improvement of 2.17 and a 0.67 reduction in integrated Brier score (IBS), often matching or surpassing the performance of models trained on real data, all while preserving privacy guarantees without significant degradation.

Abstract

Survival analysis models time-to-event data, but in clinical settings training data are costly and scarce: events accrue over years of follow-up, cohorts are small, and privacy regulations restrict sharing across institutions. Tabular generative models promise augmentation and privacy-preserving cohort sharing, yet are themselves data-hungry -- on the small cohorts typical of survival analysis, a single generator rarely characterizes the population well enough for downstream models trained on its output to match real-data performance. FoGS (Filtered Mixture-of-Generators for Survival analysis) reframes synthetic-data construction as sample selection rather than generation. A candidate pool is drawn from four architecturally distinct tabular generators, and each sample is scored by an ensemble of seven survival models trained on real data, using proper scoring rules as a per-sample plausibility proxy. A two-level pipeline optimizes, in its outer loop, a selection policy -- generator quotas, scorer weights, a random complement, and stratified balancing on event time and censoring -- against held-out downstream performance, while an inner loop tunes the downstream model (XGBoost-Cox). On 16 public datasets under train-on-synthetic, test-on-real (C-index and IBS, $0$--$100$ scale), FoGS yields mean improvements of $+2.17$ in C-index and $+0.67$ in IBS, improving both metrics on 9 of 16 datasets and at least one on 13 (one-sided Wilcoxon $p=0.039$ and $p=0.035$). It matches or exceeds real-data training on most cohorts, with no significant change in nearest-neighbour privacy margin relative to unfiltered sampling. Sample filtering over a heterogeneous generator pool is thus a viable substitute for real-data training in privacy-restricted clinical settings.

Problem

• survival analysis
• synthetic data
• data scarcity
• privacy-preserving
• tabular generative models

Innovation

• synthetic data
• survival analysis
• sample filtering
• ensemble scoring
• privacy-preserving