Women’s Health AI Consortium flags bias risk in diagnostic algorithms

Artificial intelligence is being rapidly adopted in women’s health care, but experts warn that models trained on non‑representative data can miscalculate risk for marginalized groups, worsening existing health disparities. Analyses show that diagnostic tools in obstetrics, gynecology and oncology often lack demographic parity, leading to lower specificity and sensitivity for certain racial, socioeconomic and gender‑diverse populations.

In response, the newly formed Women’s Health AI Consortium has issued standards requiring diverse clinical validation and regular audits of AI tools. Researchers also highlight that restrictive U.S. executive orders limiting gender‑related research and the continued use of binary gender categories may deepen inequities. They call for broader engagement with structural and social factors to ensure AI supports, rather than undermines, gender equity in health outcomes.