Evaluating the Potential of Reasoning Large Language Models to Perpetuate Racial and Gender Disease Stereotypes in Health Care - Takeaways - MDSpire

Evaluating the Potential of Reasoning Large Language Models to Perpetuate Racial and Gender Disease Stereotypes in Health Care

  • By

  • Joshua J Docking

  • Lee X Li

  • Bradley D Menz

  • Stephen Bacchi

  • Ashley M Hopkins

  • Michael J Sorich

  • May 28, 2026

  • 0 min

Share

  • 1

    Large language models (LLMs) risk exacerbating health disparities by perpetuating racial and gender biases in healthcare.

  • 2

    The study evaluated two reasoning LLMs, o3-mini and DeepSeek-R1, for racial and gender biases in generated clinical content.

  • 3

    Both models showed significant misrepresentation of race and gender in clinical vignettes, with over 20% misrepresentation in many cases.

  • 4

    o3-mini and DeepSeek-R1 frequently overrepresented Black populations in stereotypically associated medical conditions.

  • 5

    The findings indicate that newer reasoning models do not improve representation and may reinforce existing biases in healthcare.

Original Source(s)

Evaluating the Potential of Reasoning Large Language Models to Perpetuate Racial and Gender Disease Stereotypes in Health Care

  • by Joshua J Docking, Lee X Li, Bradley D Menz, Stephen Bacchi, Ashley M Hopkins, Michael J Sorich

  • May 28, 2026

Related Content

The Fight for Research Integrity Needs You

Elisabeth Bik has built a career exposing problematic papers. Now she is asking analytical scientists to apply the same scrutiny to their own fields.

Coercive measures in psychiatry: consensus and controversy in Swiss professional discourse

Wastewater Study Tracks Illicit Drug Use Across England

A year-long LC-MS/MS study suggests wastewater surveillance can reveal shifts in drug consumption linked to public events, weekends, and law enforcement activity

  • May 27, 2026

  • 3 min