Artificial Intelligence1 min read

Opinion: AI depends on medical records that don’t always tell the full truth about women

By · Via STAT Health · Published by Everything Blog

In short

A comparison of clinical charts with interviews of 87 women revealed extensive gaps, such as 30 instances of patients feeling disrespected with zero documentation in their files. These documentation voids mirror broader trends, with a recent survey revealing over a third of women experienced minimized or dismissed symptoms within the past five years. In one noted instance, a patient endured a decade of pelvic pain attributed to obesity and mental health until her husband attended an appointment to advocate for her. Consequently, health systems risk deploying diagnostic AI tools that inherit and amplify these biased, incomplete narratives.

Key points

  • Health systems run the risk of training AI diagnostics on data that systematically omits…: Health systems run the risk of training AI diagnostics on data that systematically omits female patients' actual medical challenges.
  • Incomplete medical charts prevent automated transcription and chart-summarizing tools fro…: Incomplete medical charts prevent automated transcription and chart-summarizing tools from understanding the true history of patient care.
  • Medical documentation must provide a more accurate account of women's experiences before…: Medical documentation must provide a more accurate account of women's experiences before AI can safely assist with treatment planning.

Sources checked by our analysis pipeline: 1

In September, the Gallup-Pivotal Women’s Health Survey reported that 34% of U.S. women strongly or somewhat agreed that a health care provider minimized or dismissed their symptoms in the past five years. Thirty-three percent reported being misdiagnosed or unable to get a diagnosis, compared with 21% of men reporting these experiences.

Medical records tell a different story. An epidemiologist and a sociologist, we recently released a study where we paired electronic medical records with detailed interviews with 87 women. We talked to these women about their pain and illnesses, work, family, poverty, violence, and encounters with the health care system.

Fifty-three of the women described a trust-breaking encounter with a health system, but only seven of them had medical records that documented mistrust. Thirty-two women reported experiencing a misdiagnosis, but misdiagnosis was only documented for eight of these women. Thirty women reported that they felt that clinicians treated them with disrespect, but this was not documented in the records anywhere.

Consider Kate (not a real name).

For more than a decade, Kate lived with debilitating pelvic pain. Her doctors repeatedly…

Artificial Intelligence