Demographically-Conditioned Synthetic Medical Images for Bias Mitigation and Bias Detection in Disease Classifiers

2026-07-16

AI Summary

This study addresses the challenge of reliably evaluating fairness in medical image classifiers for underrepresented subgroups, where scarcity of real-world test samples often undermines bias detection. The authors propose a demographic-conditioned synthetic image generator to mitigate bias during training via sequential pretraining and to construct synthetic cohorts of minority groups for precise subgroup performance auditing during evaluation. This work is the first to jointly employ conditional generative models for both bias mitigation and detection, demonstrating that synthetic data used in pretraining outperforms joint augmentation strategies. A fine-tuned Stable Diffusion 2.1 model achieves superior performance using only approximately 1% of real data volume compared to a full real-data baseline. Across five synthetic minority subgroups, the ranking of subgroup performance matches that of a high-powered real-data benchmark exactly (Spearman ρ = 1.00), substantially enhancing evaluation reliability in data-scarce scenarios.

Abstract

Per-subgroup fairness audits of medical image classifiers face a sample-size problem: minority subgroups in held-out test sets have so few samples that the resulting confidence intervals on per-subgroup performance are wider than the bias the audit is meant to detect. We argue that a demographically-conditioned synthetic generator can do both: mitigate bias on the training side and detect bias on the evaluation side. Working on COVID-19 chest CT classification with an end-to-end fine-tuned Stable Diffusion 2.1 generator, we make two findings. For bias mitigation (training), a demographically-balanced synthetic cohort is most useful as a pretraining prior, not as joint augmentation: with the same fixed data, sequential pretraining followed by fine-tuning substantially outperforms joint augmentation, and the resulting classifier surpasses the full-real baseline at $\sim$$100\times$ real-data efficiency. For bias detection (evaluation), across five synthetic minority cohorts and five classifier seeds, the synthetic estimator reproduces the subgroup ranking of a well-powered real oracle (Spearman $ρ= 1.00$ on MCC and Recall) and gives the more reliable per-cell estimate where the small real test set runs out of samples. The synthetic cohort is therefore most useful in exactly the cells that fairness audits care about, as both a fix for and a measure of subgroup bias.

Problem

• bias detection
• medical image classification
• subgroup fairness
• sample-size problem
• demographic bias

Innovation

• demographically-conditioned synthesis
• bias mitigation
• bias detection
• synthetic medical images
• fairness auditing