Demographic Bias of Expert-Level Vision-Language Foundation Models in Medical Imaging

TL;DR

This paper investigates demographic biases in state-of-the-art vision-language models for medical imaging, revealing underdiagnosis of marginalized groups and highlighting ethical concerns for equitable healthcare.

Contribution

It provides a comprehensive analysis of demographic biases in foundation models for chest X-ray diagnosis across multiple datasets, emphasizing their potential to worsen healthcare disparities.

Findings

Models underdiagnose marginalized groups compared to radiologists.

Biases are consistent across various pathologies and demographics.

Model embeddings encode significant demographic information.

Abstract

Advances in artificial intelligence (AI) have achieved expert-level performance in medical imaging applications. Notably, self-supervised vision-language foundation models can detect a broad spectrum of pathologies without relying on explicit training annotations. However, it is crucial to ensure that these AI models do not mirror or amplify human biases, thereby disadvantaging historically marginalized groups such as females or Black patients. The manifestation of such biases could systematically delay essential medical care for certain patient subgroups. In this study, we investigate the algorithmic fairness of state-of-the-art vision-language foundation models in chest X-ray diagnosis across five globally-sourced datasets. Our findings reveal that compared to board-certified radiologists, these foundation models consistently underdiagnose marginalized groups, with even higher rates seen in intersectional subgroups, such as Black female patients. Such demographic biases present over a wide range of pathologies and demographic attributes. Further analysis of the model embedding uncovers its significant encoding of demographic information. Deploying AI systems with these biases in medical imaging can intensify pre-existing care disparities, posing potential challenges to equitable healthcare access and raising ethical questions about their clinical application.