Artificial intelligence in the emergency room is consistently telling women they aren't as sick as men, even when their symptoms are identical.
Language models used for medical triage often replicate or even increase the gender biases found in human doctors. These AI systems frequently undertriage female patients, leading to longer wait times and more dangerous health outcomes. We are told that AI will remove human prejudice and make medicine more objective. This audit shows that the machines are simply digitizing and automating old stereotypes about female pain. It creates a future where a computer's unbiased decision could lead to a woman being denied life-saving care.
EQUITRIAGE: A Fairness Audit of Gender Bias in LLM-Based Emergency Department Triage
arXiv · 2605.03998
Emergency department triage assigns patients an acuity score that determines treatment priority, and clinical evidence documents persistent gender disparities in human acuity assessment. As hospitals pilot large language models (LLMs) as triage decision support, a critical question is whether these models reproduce or mitigate known biases. We present EQUITRIAGE, a fairness audit of LLM-based ESI assignment evaluating five models (Gemini-3-Flash, Nemotron-3-Super, DeepSeek-V3.1, Mistral-Small-3.