Artificial intelligence system for predicting areal bone mineral density from plain X-rays
Authors' conclusion
These findings indicate that AI can accurately predict aBMD and identify individuals at high risk of fractures.
Study context
- Population
- Vietnam Osteoporosis Study participants aged 18 years and older7,060 radiographs
- Participants
- 3,783 participants
- Comparison
- Artificial intelligence (AI)-predicted BMD from plain X-rays vs DXA-measured areal bone mineral density (aBMD)
- Outcome
- Correlation between xBMD and aBMD (Pearson's correlation coefficients)
Methodological detailsprospective population-based, ensemble of seven deep-learning models
- Prospective population-based
- Ensemble of seven deep-learning models
Performance
As reported in the abstract
Pearson r, femoral neck
0.90
95% CI 0.88–0.91
xBMD vs DXA
Pearson r, lumbar spine
0.87
95% CI 0.85–0.88
xBMD vs DXA
Further findings
AI system identification of individuals at high risk for hip fractures (AUC)
- AUC at femoral neck
- 0.96 (95% CI, 0.95-0.98)
- AUC at lumbar spine
- 0.97 (95% CI, 0.96-0.99)
Original publication
Artificial intelligence system for predicting areal bone mineral density from plain X-rays. Osteoporos Int. Published 27 August 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →