When it comes to hip fractures, time is of the essence.
Artificial Intelligence could help, suggests a recent study. When researchers pitted machine learning against human radiologists, the computer won, classifying hip fractures 19 percent more accurately than human experts.
The study, published in Nature Scientific Reports, was conducted in the United Kingdom. Like the United States, it has an aging population, and hip fractures rise along with age. There are an estimated 300,000 hip fractures every year in the United States, and that number is expected to rise to more than 500,000 by 2040.
Researchers had a minimum of two clinicians classify over 3,600 hip radiographs. But they were no match for a pair of computer models trained to do the same task. The algorithms located hip joints with overwhelming accuracy, and showed what researchers call “an impressive, and potentially significant” ability to classify the fractures.
The algorithms’ accuracy varied depending on the type of fracture, but overall their diagnoses were accurate 92 percent of the time compared with 77.5 percent of the time for the clinicians.
The researchers say their new algorithm could clear up the U.K.’s huge radiology bottleneck. Like U.S. radiologists, those clinicians simply have more work than they can complete quickly.
“This new technique we’ve shared has great potential,” said Richie Gill, a co-author of the paper who is co-director of the Centre for Therapeutic Innovation and the Institute for Mathematical Innovation at the University of Bath, in a news release. The method could achieve greater access and speed diagnoses, he said.
AI is increasingly used to beef up radiologists’ expertise. According to a 2020 study conducted by the American College of Radiology, an estimated 30 percent of radiologists use AI tools on the job, and even more are contemplating the switch.