Researchers Highlight Flaws in Medical AI's Disclosure of Race and Ethnicity Data

As artificial intelligence (AI) becomes increasingly embedded in healthcare, concerns are rising over the accuracy and transparency of race and ethnicity information in electronic health records (EHRs). Inaccurate classification of patient race and ethnicity can lead to biased AI systems, potentially harming patient care and perpetuating health disparities.

A recent publication in PLOS Digital Health emphasizes the urgent need for standardized methods for collecting race and ethnicity data. Experts in bioethics and law advocate for developers to ensure data quality and transparency, suggesting that clear disclosures about how race and ethnicity information is gathered can help stakeholders better understand the limitations and biases of AI tools.

Lead author Alexandra Tsalidis underscores the importance of transparency, comparing it to nutrition labels that inform consumers about food contents. She argues that disclosing data origins and collection methods can enhance trust and facilitate critical evaluation by regulators and patients alike.

Senior researcher Francis Shen highlights that addressing race bias in AI is crucial as these systems play a larger role in diagnosis and treatment. The proposed standards and templates aim to improve data accuracy and enable AI developers to warranty the integrity of race and ethnicity information used in their models.

Co-author Lakshmi Bharadwaj points out that fostering open dialogue and implementing best practices can significantly improve data quality. While current efforts are just the beginning, these initiatives pave the way for more equitable and trustworthy medical AI tools.

This research underscores the importance of rigorous standards in the collection and reporting of race and ethnicity data to ensure fairness and accountability in healthcare AI applications.

Source: https://medicalxpress.com/news/2025-06-medical-ai-disclose-inaccurate-ethnicity.html