Medical records often miss women's health experiences, limiting AI's potential

Ann Hirsch and Jennifer Silva argue that artificial intelligence cannot reach its full promise in women's health until medical records capture the complete picture of patients' lives. They cite a recent Gallup-Pivotal Women’s Health Survey showing that over a third of U.S. women feel their symptoms have been minimized or dismissed by providers, and many report misdiagnoses.

The authors conducted a study of 87 women, pairing electronic medical records with in‑depth interviews about pain, illness, work, family, poverty, violence, and health‑care interactions. While 53 participants described a trust‑breaking encounter, only seven had any documentation of mistrust in their records. Similarly, 32 reported misdiagnoses but only eight were recorded, and 30 felt disrespected by clinicians yet none of those experiences appeared in the charts.

These gaps suggest that current clinical data omit critical information about women's health experiences, limiting the ability of AI models to learn from real‑world outcomes. Hirsch and Silva call for more comprehensive documentation to ensure AI tools can improve diagnosis and treatment for women.