Clinical utility of nutrition and lipid metabolism in predicting bone health in patients with type 2 diabetes

BACKGROUND Bone health deterioration (BHD) is a common but often overlooked complication in patients with type 2 diabetes mellitus (T2DM), and serum biomarkers may facilitate the prediction of BHD.Nutritional and metabolic status, including lipid and hemoglobin (HGB) levels, may influence bone remodeling and provide valuable predictive information.AIM To develop and validate a screening model for prevalent BHD risk in patients with T2DM and to evaluate the contribution of nutritional and metabolic parameters.METHODS This retrospective study was conducted at Suqian First Hospital from July 2021 to December 2025 and included patients with T2DM.An independent external validation cohort was constructed from the 2017-2018 National Health and Nutrition Examination Survey (NHANES) cycle.Eight machine learning algorithms were employed to predict BHD, integrating different combinations of features such as demographics, laboratory parameters, diabetic complications, and derived biomarkers such as triglycerideglucose index combined with body mass index (TyG-BMI).Model performance was assessed and compared using multiple metrics including the area under the receiver operating characteristic curve (AUC), Brier score, and decision curve analysis.SHapley Additive exPlanations analysis was applied for model interpretability.RESULTS A total of 1022 patients were eligible for participation.In the internal validation set (n = 306), the model achieved an AUC of 0.915.In the external NHANES validation cohort (n 4 / 35 = 350), the model achieved an AUC of 0.693.The predictive factors for BHD included nutritional-metabolic biomarkers (high-density lipoprotein cholesterol, TyG-BMI, alkaline phosphatase, and HGB) and demographic variables (age and sex).An online calculator based on the BHD model was developed at https://t2d-bone- health.shinyapps.io/myapp/.CONCLUSION Nutrition and metabolism are significantly associated with bone health in T2DM patients.In this study, the categorical boosting model provided a preliminary screening tool to identify individuals with existing BHD who may benefit from targeted nutritional assessments and lifestyle interventions.Further recalibration, multicenter prospective validation and clinical utility evaluation are needed prior to clinical adoption.