Fracture risk prediction in postmenopausal women with traditional and machine learning models in a nationwide, prospective cohort study in Switzerland with validation in the UK Biobank
Authors' conclusion
We conclude that fracture risk can be improved by including the lumbar spine T-score, trabecular bone score, numbers of falls and recent fractures, and treatment information has a significant impact on fracture prediction.
Study context
- Population
- Postmenopausal women in a nationwide Swiss Osteoporosis Registry (January 2015–October 2022), validated in UK Biobank
- Participants
- 5,944 vs 5,474
- Exposure
- Traditional and machine learning fracture prediction models (Random survival forest, eXtreme Gradient Boosting) vs FRAX Switzerland
- Swiss Osteoporosis Registryn = 5,944
- UK Biobankn = 5,474
- Outcome
- Prediction of vertebral, hip and any fractures (Uno's C-index, time-dependent AUC)
Methodological detailsnationwide, traditional and machine learning survival models, independent validation in UK Biobank
- Nationwide
- Traditional and machine learning survival models
- Independent validation in UK Biobank
Main finding
As reported in the abstract
C-index in the independent validation set · year 2
C-index, vertebral fractures
0.74
0.58, 0.86
C-index, hip fractures
0.83
0.7, 0.94
C-index, any fractures
0.63
0.58, 0.69
Observational data: associations, not effects.
Further findings
FRAX Switzerland 10-yr fracture probability
- Major osteoporotic fractures
- 0.60 [0.55, 0.64]
- Hip fractures
- 0.62 [0.49, 0.74]
Most important variables (Shapley additive explanations)
- Age, T-scores and prior fractures; number of falls important predictor of hip fractures
Traditional vs machine learning models
- Similar C-indices
Original publication
Fracture risk prediction in postmenopausal women with traditional and machine learning models in a nationwide, prospective cohort study in Switzerland with validation in the UK Biobank. J Bone Miner Res. Published 5 June 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →