Establish and validate the reliability of predictive models in bone mineral density by deep learning as examination tool for women
Authors' conclusion
While FRAX with BMD could be more precise in estimating the fracture risk, DL-based models were validated to slightly reduce the number of under- and over-treated patients when no BMD measurements were available.
Study context
- Population
- Women enrolled from the National Bone Health Screening Project of Taiwan (mean age 65.4 ± 9.4 years)
- Participants
- 10,827 women
- Exposure
- Deep learning (DL)-based classification and FRAX without bone mineral density (BMD) vs FRAX with BMD
- Outcome
- Performance of DL-based models in predicting BMD/lowest T-score and classifying patients at high fracture risk vs FRAX with BMD
Methodological detailsdeep learning-based predictive models, Bland-Altman analysis
- Deep learning-based predictive models
- Bland-Altman analysis
Main finding
As reported in the abstract
Classification of high fracture risk vs FRAX with BMD (reference)
| Group | Accuracy rate | PPV | NPV |
|---|---|---|---|
| FRAX WITHOUT BMD | 78.8% | 64.6% | 89.9% |
FRAX WITHOUT BMD
Accuracy rate78.8%
PPV64.6%
NPV89.9%
Observational data: associations, not effects.
Further findings
Prediction of BMD and lowest T-score (root-mean-square error (RMSE))
- RMSE lumbar spine BMD
- 0.099
- RMSE total hip BMD
- 0.089
- RMSE femoral neck BMD
- 0.076
- RMSE lowest T-score
- 0.68
- Bland-Altman: nonsignificant difference between predictive models and DXA
FRAX score with femoral neck BMD
- Major osteoporotic fracture risk
- 9.7% ± 6.7%
- Hip fracture risk
- 3.3% ± 4.6%
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Establish and validate the reliability of predictive models in bone mineral density by deep learning as examination tool for women. Osteoporos Int. Published 20 September 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →