Sjögren's disease & overlap diseases
Osteoporosis prediction in primary Sjögren’s syndrome: development and external validation of a machine-learning comparison model
Authors' conclusion
A transparent, externally validated seven-variable model provides individualized DXA-defined osteoporosis risk estimation in pSS
Study context
- Population
- Patients with primary Sjögren's syndrome (pSS), Longhua Hospital, with external validation cohort from Shanghai Seventh People's Hospitalplus 266 external validation
- Participants
- 1,000 patients
- Exposure
- Logistic regression vs seven supervised machine-learning algorithms
- External validationn = 266
- Outcome
- DXA-defined osteoporosis (model performance by AUC, calibration and decision curve analysis)
Methodological detailsdevelopment and external validation of a prediction model, random 7:3 training/internal validation split, logistic regression compared with seven supervised machine-learning algorithms, Freely accessible web-based calculator developed for point-of-care use
- Development and external validation of a prediction model
- Random 7:3 training/internal validation split
- Logistic regression compared with seven supervised machine-learning algorithms
- Freely accessible web-based calculator developed for point-of-care use
Main finding
As reported in the abstract
AUC of final seven-predictor logistic regression model
| Group | AUC |
|---|---|
| Training | 0.820 |
| Internal validation | 0.807 |
| External validation | 0.787 |
AUC
Training0.820
Internal validation0.807
External validation0.787
Observational data: associations, not effects.
Further findings
Machine-learning models vs logistic regression
- Machine-learning models achieved higher training AUCs but showed poorer transportability
Original publication
Osteoporosis prediction in primary Sjögren’s syndrome: development and external validation of a machine-learning comparison model. Rheumatology (Oxford). Published 19 August 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →