“A better understanding of how age and comorbid conditions affect mammography use may help target specific populations and improve use of preventive care," wrote Cindy Yuan, MD, PhD, with the University of Chicago’s Department of Radiology, and colleagues.

Trained on nearly 1 million screening mammography images, researchers from New York University found their algorithm could push radiologists’ ability to accurately identify breast cancer to nearly 90%.

"We've been telling patients that even if these injections don't relieve your pain, they're not going to hurt you," said study author Ali Guermazi, MD, PhD. "But now we suspect that this is not necessarily the case."

"The AI we use works almost like magic—and it will help radiologists save lives," said Antonio Sze-To, a postdoctoral fellow who worked on the project.

Overall rates of low-value diagnostic cardiovascular testing—including imaging—have dropped over the past 10 years, while rates of high-value, guideline-recommended testing rose slightly.

The findings were true regardless of tumor type, size or cancer stage, wrote Pragya A. Dang, MD, and colleagues at Brigham and Women’s Hospital in Boston.