Acquisition parameters influence AI recognition of race in chest x-rays and mitigating these factors reduces underdiagnosis bias

• Published in: Nature communications Vol. 15; no. 1; pp. 7465 - 11
• Main Author: Lotter, William
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 29-08-2024; Nature Publishing Group; Nature Portfolio
• Subjects: 639/705/1042; 639/705/117; 692/700/1421; 692/700/3935; Adult; Artificial Intelligence; Bias; Chest; Data acquisition; Datasets; Demographics; Demography; Female; Health care; Healthcare Disparities; Humanities and Social Sciences; Humans; Image acquisition; Male; Medical imaging; Middle Aged; multidisciplinary; Parameter identification; Patients; Performance prediction; Population studies; Process parameters; Racial Groups; Radiography, Thoracic; Science; Science (multidisciplinary); X-rays
• ISSN: 2041-1723; 2041-1723
• DOI: 10.1038/s41467-024-52003-3

A core motivation for the use of artificial intelligence (AI) in medicine is to reduce existing healthcare disparities. Yet, recent studies have demonstrated two distinct findings: (1) AI models can show performance biases in underserved populations, and (2) these same models can be directly trained to recognize patient demographics, such as predicting self-reported race from medical images alone. Here, we investigate how these findings may be related, with an end goal of reducing a previously identified underdiagnosis bias. Using two popular chest x-ray datasets, we first demonstrate that technical parameters related to image acquisition and processing influence AI models trained to predict patient race, where these results partly reflect underlying biases in the original clinical datasets. We then find that mitigating the observed differences through a demographics-independent calibration strategy reduces the previously identified bias. While many factors likely contribute to AI bias and demographics prediction, these results highlight the importance of carefully considering data acquisition and processing parameters in AI development and healthcare equity more broadly.
Artificial intelligence (AI) models can perform unequally across patient groups and can also be trained to recognize patient demographics. Here, the authors show that image acquisition parameters influence AI models trained to predict patient race from chest x-rays and that mitigating these factors reduces an underdiagnosis bias.