Machine learning-based screening model for osteoporosis in Chinese postmenopausal women: a nationwide study with prospective fracture validation
Authors' conclusion
This model demonstrated good performance in predicting osteoporosis among Chinese postmenopausal women and may facilitate two-step screening strategies.
Study context
- Population
- Chinese postmenopausal women
- Participants
- 6,574 / 1,758 / 1,439 / 795
- Comparison
- Training vs China Vertebral and Osteoporosis Study vs Peking Vertebral Fracture Study baseline vs 5-year follow-up
- Trainingn = 6,574
- China Vertebral and Osteoporosis Studyn = 1,758
- Peking Vertebral Fracture Study baselinen = 1,439
- 5-year follow-upn = 795
- Outcome
- Osteoporosis prediction (AUC and calibration)
Methodological detailsmachine learning, LASSO-selected predictors, external validation, prospective fracture validation, nationwide, A web-based calculator was developed for public use
- Machine learning
- LASSO-selected predictors
- External validation
- Prospective fracture validation
- Nationwide
- A web-based calculator was developed for public use
Performance
As reported in the abstract
Discrimination for osteoporosis (AUC)
| Test or model | AUC |
|---|---|
| Training | 0.798 |
| China Vertebral and Osteoporosis Study | 0.750 |
| Peking Vertebral Fracture Study baseline | 0.744 |
AUC
Training0.798
China Vertebral and Osteoporosis Study0.750
Peking Vertebral Fracture Study baseline0.744
- AUC in the internal validation cohort
- 0.775
Further findings
Clinical fractures, cumulative incidence, high risk vs other · 5 years
| HR (95% CI) | P value | ||
|---|---|---|---|
| High risk | 1.78 (1.06–2.99) | P = 0.026 | |
| 0.20.5125Hazard ratio (log scale) |
| Outcome | P value |
|---|---|
| All fractures, high risk vs other · 5 years15.9% vs. 9.5% | P = 0.015 |
| Non-traumatic fractures, high risk vs other · 5 years7.8% vs. 3.9% | P = 0.041 |
Rule-out threshold (≥ 0.098) in two external cohorts
- Sensitivity
- 94.7%, 89.3%
- NPV
- 91.3%, 92.7%
Original publication
Machine learning-based screening model for osteoporosis in Chinese postmenopausal women: a nationwide study with prospective fracture validation. Osteoporos Int. Published 27 August 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →