Unless designers quickly change their approach, new consumer technologies could worsen disparities in healthcare, activists say.
Many people are unable to pay for devices that can monitor their health conditions, or don’t have good internet connections, or don’t have the education necessary to make use of the devices, according to leaders of advocacy organizations invited to the Consumer Electronics Show (CES) 2022 Annual Meeting
At the same time, inventors are building their racial, gender, age, and economic biases into the devices and into the artificial intelligence (AI) that analyzes the data the devices produce, the activists say.
“Pulse oximeter, blood pressure monitors, glucometers for people with diabetes, those things are still pretty much out of reach for a lot of these folks who can really benefit from them upstream,” said Lisa Fitzpatrick, MD, MPH, MPA, founder and CEO of Grapevine Health, a Washington, DC–based organization that promotes healthcare literacy among underserved people. “I think it will exacerbate the disparities if the access is uneven.”
Many companies building the new devices don’t see patients served by Medicaid or Medicare as an important part of their market, she said.
“A lot of these companies are doing just fine building for high-income populations and privately insured patient populations,” she said. “If they’re making a good profit, I don’t know how invested or motivated they are to try and get these other populations.”
The design of the devices sometimes doesn’t take cultural differences into account, either, she said. For example, a voice recognition system created by a White person might recognize the sentence, “I have diabetes,” and not recognize the sentence, “I have sugar,” which some Black people use to express the same thing.
Meanwhile, White men are overrepresented in most datasets used to train AI systems, says Austin Cason, founder and CEO of Washington, DC–based SeedAI, which is working to bring AI to underserved groups. “You’re going to miss things because the tools aren’t going to work as well.”
This can have important consequences for programs used in diagnosis. For example, a study published in Computer Vision and Pattern Recognition found that light-skin images predominate in dermatology atlases and that the accuracy of AI systems trained to recognize skin conditions varies depending on skin type.
Cason called on AI developers to partner with diverse institutions such as community colleges and historically black universities and colleges. The National Science Foundation has started to fund such efforts, he said.
Annie Hardy, senior visioneer at Cisco, agreed that healthcare AI systems will exacerbate disparities if designed poorly. But they also have the potential to reduce disparities if designed well, she said.
She gave the example of systems trained to recognize pain by analyzing human speech. A White male physician might be less likely to recognize the distress of a woman of color, but a properly trained AI system would. “It can inform doctors or staff about information that is similar to other cases where a mother died in childbirth ― how she’s feeling, what she’s saying, what some of the words are that they may not catch.”
Although many consumer health monitoring companies pay lip service to the importance of reducing disparities, few have actually worked with public payers or created partnerships with underserved communities, Fitzpatrick said.
She advised designers to get suggestions from patients and providers at community clinics and federally qualified health centers.
Companies producing AI also need to take on the burden of training more diverse programmers rather than relying on graduates of elite universities, said Ximena Hartsock, PhD, founder of Phone2Action, who moderated a CES panel on diversity in artificial intelligence. “There is a huge opportunity in using apprenticeships,” she said.
Consumer Electronics Show (CES) 2022 Annual Meeting.
Laird Harrison writes about science, health, and culture. His work has appeared in magazines, newspapers, on public radio, and on websites. He is at work on a novel about alternate realities in physics. Harrison teaches writing at the Writers Grotto. Visit him at lairdharrison.com or follow him on Twitter: @LairdH.
Content Source: https://www.medscape.com/viewarticle/966404?src=rss