Jeff Stringer, division director of global women’s health at the UNC School of Medicine, created a low-cost, AI-assisted ultrasound device with his team that is able to diagnose gestational age, detect twins, estimate fetal weight and identify breech presentation, all from devices like phones, laptops or iPads.
Stringer said concerns about bias do not affect his work, because the data used to train the AI tools comes from a small, specific data set. Because these models are focused on specific diagnoses, the AI models are highly accurate. However, he said it is well-known that much more scientific research has been done on men than on women, and developers have to be careful to make sure models don’t give answers with a male bias.
Lamm said to address some of these concerns, UNC Health has been careful not to use fully autonomous AI and have made sure that when they use AI as a tool, there is still a human in control.
Lamm added that UNC Health has a professionally diverse governance group and operational leaders that review every AI capability the hospital is considering to ensure that it is safe and ethical.
As the use of AI tools in healthcare has increased, so have calls for legislative oversight.
Compared to other states, North Carolina has not done much in terms of regulating the use of AI tools in healthcare.
Stringer said he supports legislation that provides clear guidance on the safe use of AI in healthcare, as long as it is done with the right intent. He said AI should be regulated in a deliberate and responsible, but also cautious way.
“Given the situation of an aging population and an overburdened and overworked workforce, I think that it’s really important that we figure out how to do AI right,” Silcox said. “And doing AI right is not obvious, and it’s not easy, but I do think it is going to be important and have enormous benefits.”
@DTHCityState | city@dailytarheel.com
To get the day’s news and headlines in your inbox each morning, sign up for our email newsletters.