Researchers Call for Standardization of Race and Ethnicity Data in Medical AI to Address Bias

The Challenge of Inaccurate Race and Ethnicity Data in Medical AI

A new study published in PLOS Digital Health has shed light on a critical issue in the rapidly evolving field of medical artificial intelligence (AI). Researchers have identified that inaccurate race and ethnicity data in electronic health records (EHRs) could significantly impact patient care as AI becomes more integrated into healthcare systems 1

1

.

The problem stems from inconsistent data collection methods and difficulties in accurately classifying individual patients' race and ethnicity. As a result, AI systems trained on these datasets risk inheriting and perpetuating racial bias, potentially compromising the quality and equity of healthcare delivery.

Proposed Solutions and Best Practices

To address this pressing concern, the research team, led by Alexandra Tsalidis, MBE, has called for immediate action on two fronts:

1. Standardization of data collection: The study emphasizes the need for healthcare systems to adopt standardized methods for collecting race and ethnicity data 2

   2

   .

2. Data quality warranty: The researchers propose that AI developers should provide warranties for the quality of race and ethnicity data used in their medical AI systems.

These recommendations aim to improve data accuracy and transparency in medical AI development. The research team has also developed a new template for AI developers to transparently warrant the quality of their race and ethnicity data.

Transparency and Consumer Safety

Lead author Alexandra Tsalidis draws an analogy between the proposed data quality disclosures and nutrition labels on food products. She states, "Just as nutrition labels inform consumers about what they're putting into their bodies, these disclaimers can reveal the quality and origins of the data used to train AI-based health care tools" 1

1

.

This approach is expected to not only advance transparency in medical AI but also empower patients and regulators to critically assess the safety of AI-based medical devices.

The Importance of Addressing Racial Bias in AI

Senior author Francis Shen, JD, PhD, emphasizes the significance of this research, stating, "Race bias in AI models is a huge concern as the technology is increasingly integrated into healthcare. This article provides a concrete method that can be implemented to help address these concerns" 2

2

.

The study offers a starting point for tackling this complex issue. Co-author Lakshmi Bharadwaj, MBE, suggests that an open dialogue regarding best practices is crucial, and the proposed approaches could lead to significant improvements in the field.

Research Support and Future Directions

This important research was supported by the NIH Bridge to Artificial Intelligence (Bridge2AI) program and an NIH BRAIN Neuroethics grant (R01MH134144) 1

1

. While the study provides a solid foundation for addressing racial bias in medical AI, the authors acknowledge that more work needs to be done in this area.

As AI continues to play an increasingly significant role in healthcare, ensuring the accuracy and fairness of these systems becomes paramount. The standardization of race and ethnicity data collection and the implementation of data quality warranties represent crucial steps towards more equitable and reliable AI-driven healthcare solutions.