Development of a Predictive Model and Nomogram for Neuropathy Risk in T2DM Patients

The global prevalence of type 2 diabetes mellitus (T2DM) continues to rise, with diabetic neuropathy significantly impacting patient outcomes. This study developed a predictive nomogram for neuropathy risk in T2DM patients using retrospectively analyzed electronic medical records (2013–2023) from the Affiliated Hospital of North Sichuan Medical College. After rigorous data cleaning, univariate logistic regression and XGBoost screening identified ten predictors including age, creatine kinase (CK), total urinary protein, free thyroxine (FT4), α-hydroxybutyrate dehydrogenase (α-HBDH), cystatin C (CysC), urinary creatinine (Ucr), serum calcium (Ca), α1-microglobulin (X1-MG), and urinary albumin (UALB), integrated via multivariate logistic regression into a visual nomogram. These variables were integrated into a visual nomogram via multivariate logistic regression. Model evaluation demonstrated robust discriminative ability, with area under the receiver operating characteristic curve (AUC) values of 0.743 (95% CI: 0.733–0.754) in the training dataset and 0.751 (95% CI: 0.735–0.768) in the testing dataset. Calibration curves confirmed prediction consistency, while DCA highlighted significant clinical net benefit. This nomogram provides a practical, visual tool for clinicians to estimate individual neuropathy risk, enabling early identification of high-risk patients and targeted interventions to delay or prevent neuropathy onset. The model demonstrates reliable predictive accuracy and substantial clinical utility in managing T2DM complications.