Objective <p>Diabetic tibial nerve neuropathy (DTN), a severe subtype of diabetic peripheral neuropathy, is often underdiagnosed in the early stages. This significantly raises the risks of foot ulcers and amputations. This study aims to develop and validate a novel nomogram prediction model integrating clinical characteristics and ultrasound radiomics features for early identification of DTN patients.</p> Methods <p>From January 2024 to April 2025, 300 patients with type 2 diabetes who visited the Second Affiliated Hospital of Fujian Medical University were prospectively collected and randomly divided into training and validation cohorts at a ratio of 7:3. The study collected 134 patients from two other tertiary hospitals as the external validation set. The differences among the three cohorts were evaluated by single-factor ANOVA analysis of variance or χ² test. Tibial nerve ultrasound cross-sectional images were analyzed to extract radiomics features. Optimal features were selected using t-tests and least absolute shrinkage and selection operator (LASSO) regression, generating a weighted ultrasound radiomics score (Rad-score). A nomogram integrating clinical variables and Rad-score was developed through univariate and multivariate logistic regression. Performance was evaluated using AUC, calibration curves, decision/clinical impact curves.</p> Results <p>The nomogram incorporated three clinical variables (age, smoking history, and hypertension history) and Rad-score derived from seven optimal ultrasound radiomics features. The model demonstrated AUCs of 0.947 (training set), 0.910 (internal validation set) and 0.887 (external validation set). Calibration and decision curves indicated strong consistency and clinical utility.</p> Conclusion <p>The clinical-ultrasound radiomics nomogram effectively predicts DTN risk, potentially serving as a reference tool for early diagnosis.</p>