STAT+: U.K. researchers warn about biases in medical devices, health care algorithms
Regulators, clinicians, and health care algorithm developers need to take additional steps to ensure that medical devices work avoid hidden biases, U.K. researchers warned.
LONDON — Regulators, clinicians, and health care algorithm developers need to take additional steps to ensure that medical devices work equally well for all patients, avoiding blind spots that can lead to worse care for patients from underrepresented racial and ethnic groups, according to a U.K.-commissioned review released on Monday.
The report, which also warned about the products’ biases potentially hurting women and people from lower socioeconomic groups, called on the government to improve its understanding of the devices used commonly in the country’s health service, with a need for an expert panel that can assess the possible unintended consequences as AI tools expand. Providers also need to learn about the limitations of these devices, to ensure they don’t result in poor patient care.
The report cited several examples, including evidence that pulse oximeters, which track blood oxygen levels, can overestimate such levels in patients with darker skin, potentially providing reassurance instead of indicating they need to be treated.
What's Your Reaction?
