Development and validation of common data model-based fracture prediction model using machine learning algorithm
Authors' conclusion
We developed scoring systems for osteoporotic fractures from hospital-based cohorts and validated them in an independent cohort.
Study context
- Population
- Adults aged ≥ 50 years with BMD data by DXA, hospital-based cohorts in Korea (mean age 64.5 years; 84.3% women)mean follow-up 7.6 years
- Participants
- 20,107 vs 13,353
- Comparison
- Discovery vs Validation
- Discoveryn = 20,107
- Validationn = 13,353
- Outcome
- Major osteoporotic and hip fracture events
Methodological detailscommon data model database, DeepHit and Cox proportional hazard models, independent validation cohort, hospital-based cohorts
- Common data model database
- DeepHit and Cox proportional hazard models
- Independent validation cohort
- Hospital-based cohorts
Performance
As reported in the abstract
Harrell's C-index of the scoring models
| Test or model | C-index, major osteoporotic fractures | C-index, hip fractures |
|---|---|---|
| Discovery | 0.789 | 0.860 |
| Validation | 0.762 | 0.773 |
C-index, major osteoporotic fractures
Discovery0.789
Validation0.762
C-index, hip fractures
Discovery0.860
Validation0.773
- Major osteoporotic fracture events observed
- 1990
- Hip fracture events observed
- 309
Further findings
Estimated 10-year fracture risk by score
- Major osteoporotic fracture risk at score 0
- 2.0%
- Hip fracture risk at score 0
- 0.2%
- Major osteoporotic fracture risk at maximum score
- 68.8%
- Hip fracture risk at maximum score
- 18.8%
Original publication
Development and validation of common data model-based fracture prediction model using machine learning algorithm. Osteoporos Int. Published 17 May 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →