Retrospective Development and Internal Validation of a Diagnostic Model to Identify Secondary Osteoporosis in Community-Dwelling Older Men

Development and validation of a robust logistic regression model for predicting osteoporosis in older people

Background Osteoporosis (OP) represents a significant public health challenge in the aging population, often leading to debilitating fractures. Early identification of high-risk individuals through routine clinical data is vital for preventive care. This study aimed to develop and validate a robust machine learning-based framework to predict OP in older people using comprehensive physical examination indicators. Methods We analyzed a retrospective single-center cohort of 852 participants (age ≧ 60 years) from a physical examination center. The cohort was randomly partitioned into a training set (n = 598) and an internal validation set (n = 254) at a 7:3 ratio. Nineteen potential predictors, including demographics (Age, Sex, BMI), lifestyle habits, comorbidities, and biochemical markers (e.g., SUA, Hb, LDL), were evaluated. After feature selection via LASSO regression, five algorithms-Logistic Regression (LR), Random Forest (RF), Decision Tree (DT), Support Vector Machine (SVM), and XGBoost-were constructed. Hyperparameter tuning was performed only within the training set using 5-fold cross-validation and grid search, and the internal validation set was kept separate for final performance assessment. Model performance was evaluated using the Area Under the Curve (AUC), calibration curves, and Decision Curve Analysis (DCA). Restricted cubic spline (RCS) analysis was further employed to evaluate dose–response associations between selected variables and osteoporosis risk. Results The prevalence of OP in the study population was 28.87% (246/852). LASSO regression identified 12 optimal features for model development. While ensemble methods (RF and XGBoost) exhibited high discriminative power, they showed signs of potential overfitting with perfect training set performance. In contrast, the LR model demonstrated superior robustness and stability, achieving an AUC of 0.809 in the training set and 0.782 in the internal validation set. In the validation cohort, the LR model maintained balanced performance with a sensitivity of 0.767 and a specificity of 0.702. Calibration curves indicated agreement between predicted and observed risks, and DCA suggested clinical net benefit across a broad range of threshold probabilities. SHAP (SHapley Additive exPlanations) analysis indicated that Sex, Age, and BMI were the most influential predictors. Lower levels of serum creatinine (Scr), serum uric acid (SUA), and fasting plasma glucose (FPG) were also associated with higher predicted osteoporosis risk. Multivariable-adjusted RCS analysis suggested an upward trend between HDL-C and OP risk, particularly in the female subgroup (OR = 3.35, 95% CI: 0.86–13.04; p = 0.081), but this finding did not reach conventional statistical significance and should be interpreted cautiously. Conclusion Compared with complex ensemble algorithms, the LR model provides a stable and interpretable approach for identifying older people who may warrant DXA assessment or closer bone-health evaluation. Because this study used a retrospective single-center cohort and internal validation only, external multi-center validation is required before routine clinical deployment.

Clinical and Laboratory Predictors of Bone Mineral Density Trajectories and Osteoporosis Progression: A Retrospective Longitudinal Cohort Study

No routinely measured biomarker demonstrated clinical utility for identifying BMD trajectory in this cohort of Saudi dual-energy X-ray absorptiometry patients, suggesting future studies with explicit treatment tracking and repeated bone-turnover measurements are required for confirmation.

Age-specific reference values for lumbar spine trabecular bone score (TBS) in Brazilian adult women.

It is indicated that trabecular microarchitecture deteriorates earlier and more rapidly with aging than bone mass loss captured by BMD, supporting the use of TBS as a complementary tool for evaluating skeletal health and fracture risk.

Prevalence of densitometric osteoporosis in women younger than 65 years without indications for DXA screening

Densitometric osteoporosis was identified in a relevant proportion of women younger than 65 years without traditional risk factors, supporting an individualized approach to bone health assessment.

Clinical characteristics of the reduced bone mineral density subgroup among adults with type 2 diabetes mellitus: a retrospective cross-sectional study

Skeletal fragility in type 2 diabetes mellitus (T2DM) is incompletely represented by areal bone mineral density (BMD). We defined an operational reduced-BMD subgroup using routine hospital information-system data and examined whether routinely available clinical variables showed incremental associations and whether complex classifiers improved internal discrimination beyond age and sex. This single-center retrospective cross-sectional analysis included 329 de-identified analytical records from adults with T2DM who underwent routine dual-energy X-ray absorptiometry. The operational reduced-BMD subgroup was defined by the lowest T-score among the lumbar spine L1-L4, femoral neck, and total hip: records with at least one site-specific T-score ≤ − 1.0 were classified into the subgroup, and a stricter < -1.0 definition was examined as a boundary sensitivity analysis. Because menopausal status and Z-scores were unavailable, this rule was treated as an operational subgroup definition rather than a universal clinical diagnosis. Descriptive analyses, deliberately parsimonious Jeffreys-prior bias-reduced logistic models, fixed-unit age sensitivity analyses, and repeated nested internal classification were performed. Twenty-six records (7.9%) met the primary operational definition and 22 met the stricter definition; no record met a T-score ≤ − 2.5 threshold at the retained primary sites. Twenty-five of the 26 records in the reduced-BMD subgroup were from women. In the prespecified full-cohort model, age had an odds ratio (OR) of 5.39 per within-sample standard deviation (95% profile-likelihood confidence interval [CI], 2.71–12.15), and the female-sex estimate was large but imprecise (OR, 34.47; 95% CI, 8.29-321.11). On a common five-year scale, the age OR was 2.33 (95% CI, 1.65–3.50) in the full cohort and 2.29 (95% CI, 1.60–3.47) among records aged ≥ 50 years. The fixed age-sex classifier had a median internally validated ROC AUC of 0.921 (range, 0.915–0.923), decreasing to 0.901 (range, 0.897–0.904) in the age ≥ 50 subset. Among the tested algorithms and prespecified features, no complex classifier materially exceeded the fixed age-sex benchmark. In this selected single-center record-level dataset, the operational reduced-BMD subgroup was characterized principally by older age and marked female predominance. Restriction to age ≥ 50 years reduced apparent discrimination because of narrower case mix, while the age association remained similar on a fixed five-year scale. The additional tested variables did not provide precisely estimated incremental associations in the prespecified age-sex-plus-one-variable models, and the tested complex classifiers did not materially improve internal discrimination over the age-sex benchmark. All classification estimates were internally validated only, and the findings do not support causal, diagnostic, prognostic, deployment, or treatment claims. The operational reduced-BMD subgroup comprised 26 of 329 analytical records from adults with T2DM. Twenty-five of the 26 reduced-BMD records were from women, and all occurred at age ≥50 years. Jeffreys-prior bias-reduced logistic regression addressed sparse outcomes and near-separation. The age association was similar in the full cohort and the age ≥50 subset when expressed per five years. Within internal validation, the tested complex classifiers did not materially outperform the age-sex benchmark.

Fragility fracture site is associated with frailty in older adults: a population-based cohort study in Japan.