Retrospective single-centre model predicting osteoporosis in people aged 60 and over
In a retrospective single-centre cohort of 852 people aged 60 and over, a logistic regression model using 12 selected features reached an AUC of 0.782 in internal validation; external validation is still required. [1]
Automatic summary · not editorially reviewed
The key findings
The retrospective single-centre cohort included 852 participants aged 60 and over, with an osteoporosis prevalence of 28.87%. [1]
The logistic regression model reached an AUC of 0.782 in the internal validation set, with sensitivity 0.767 and specificity 0.702. [1]
The HDL cholesterol trend in the female subgroup did not reach conventional statistical significance (p = 0.081). [1]
The source at a glance
- Study type / format
- Retrospective single-centre cohort study developing and internally validating prediction models.
- Population / context
- People aged 60 and over attending a physical examination centre (852 participants).
- Summary based on
- Study abstract
- Original publication
- July 28, 2026
Osteoporosis prediction in 852 older examinees
The study aimed to develop and validate a machine learning framework to predict osteoporosis in older people from routine physical examination data. It analysed a retrospective single-centre cohort of 852 participants aged 60 and over from a physical examination centre. Osteoporosis prevalence in this population was 28.87% (246 of 852). Nineteen potential predictors were evaluated, including age, sex, body mass index, lifestyle habits, comorbidities and biochemical markers.
Supporting source [1]
Logistic regression stayed stable in internal validation
The cohort was split randomly into a training set (598) and an internal validation set (254). After feature selection, 12 features were used. Ensemble methods showed high discrimination but signs of potential overfitting with perfect training performance. The logistic regression model reached an AUC of 0.809 in training and 0.782 in internal validation, with sensitivity 0.767 and specificity 0.702 in validation. Sex, age and body mass index were the most influential predictors. Lower serum creatinine, serum uric acid and fasting plasma glucose were associated with higher predicted risk.
Supporting source [1]
HDL-C trend was not statistically significant
A multivariable-adjusted analysis suggested an upward trend between HDL cholesterol and osteoporosis risk, particularly in the female subgroup (odds ratio 3.35, 95% confidence interval 0.86–13.04; p = 0.081). This did not reach conventional statistical significance and the authors advise cautious interpretation. Because the study was a retrospective single-centre cohort with internal validation only, external multi-centre validation is required before routine clinical use.
Supporting source [1]
How to interpret this
Abstract only; the full text was not reviewed. This was a retrospective single-centre cohort with internal validation only, so external multi-centre validation is needed before routine clinical deployment. The HDL cholesterol finding was not statistically significant.
Based on the study abstract; the full paper, tables and supplementary data were not reviewed.
Source & supporting passages
Xue H, Hao X, Liu D, Li Y, Yu H. · europepmc.org · July 28, 2026