Chest x-ray AI detects osteoporosis across seven countries

Tuesday, December 1 | 1:30 p.m.-2:30 p.m. | T6-SSCH05-1 | Room E451A

An AI model that analyzes routine chest x-rays for signs of reduced bone mineral density held its own across a broad multinational dataset, according to research in this session.

In a study involving 5,873 images from 5,495 patients in seven countries spanning five geographic regions, from the U.S. and Ireland to Jordan, India, Brazil, and Saudi Arabia. The tool (PROS CXR: OSTEO, Promedius) sorted images into three bone mineral density categories: normal, osteopenia, and osteoporosis.

Compared to bone scans as the reference standard, for osteoporosis detection, it achieved a pooled area under the curve of 0.819 and a sensitivity of 87%, with country-level AUC values ranging from 0.702 to 0.866. A negative predictive value of 94% suggests reliable rule-out performance across the international cohort.

The researchers designed the study to test whether the tool could support opportunistic screening in settings where DEXA is underutilized, with the goal of identifying patients at high risk for bone mineral density loss before a dedicated bone density workup is ordered.

“Such CXR-based AI tools can help address global DEXA underutilization for dedicated osteoporosis screening and flagging high-risk patients for BMD loss,” the authors wrote.

Stop by this Tuesday session to hear how performance varied country by country and what the researchers see as next steps toward clinical deployment.

Page 1 of 1