Construction and validation of a diagnostic model for type 2 diabetes mellitus comorbid with metabolic...

The prevalence of type 2 diabetes mellitus (T2DM) with concurrent Metabolic dysfunction-Associated Steatotic Liver Disease (MASLD) is high. However, simple, efficient and non-invasive diagnostic tools based on routine serological indicators are lacking. This study aims to construct and validate a simple diagnostic model using routine serological indicators.

A retrospective case-control study is conducted, enrolling 470 patients with T2DM. They are randomly stratified into a training set and a validation set at a 7:3 ratio. Independent predictive factors are identified using univariate and multivariate logistic regression. Diagnostic performance is assessed via receiver operating characteristic (ROC) curves. A nomogram diagnostic model is constructed and evaluated using ROC curves, calibration curves and decision curve analysis. For external validation, an additional 300 T2DM patients are recruited.

Body mass index (BMI), fasting C-peptide, total cholesterol, triglycerides, high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), urea and alanine aminotransferase (ALT) are independent predictors of comorbid T2DM and MASLD. The area under the curve (AUC) of this model in the training set, internal validation set and external validation set is 0.921, 0.913 and 0.830, respectively. The calibration curve closely approximates the ideal diagonal line and decision curve analysis (DCA) demonstrates significant clinical net benefit across a broad range of threshold probabilities.

This model enables high-precision diagnosis of T2DM with MASLD using routinely available serological indicators, avoiding invasive procedures and exhibiting strong potential for clinical translation. Future studies should validate its generalizability in multicenter prospective cohorts to promote its use as a routine screening tool for MASLD in T2DM patients.