Abstract: Objective To establish and validate a nomogram model based on multimodal ultrasound radiomics to predict the risk of cardiovascular diseases （CVD） in patients with diabetic kidney disease （DKD）. Methods A retrospective analysis of baseline clinical data and multimodal ultrasound images from 167 DKD patients was conducted. The ultrasound images were segmented and radiomic features were extracted using 3D Slicer software. A radiomics score was generated for each patient using the least absolute shrinkage and selection operator （LASSO） algorithm. Based on these scores and clinical predictors related to CVD, a nomogram model was established to predict the risk of CVD in DKD patients. The model was internally validated using the receiver operating characteristic （ROC） curve, calibration curve, and decision curve analysis （DCA）. Results Utilizing the LASSO algorithm, five radiomic features closely related to the incidence of CVD were extracted from the dual mode elastography of each patient. The radiomics score, which was constructed based on these features, predicted the risk of CVD with an area under ROC curve （AUC） of 0.764. In terms of clinical features, age, body mass index, low-density lipoprotein cholesterol, and glycated hemoglobin were identified as independent risk factors for developing CVD. A radiomics nomogram, built based on these factors and the radiomics score, significantly outperformed the radiomics score alone in predicting the risk of CVD, with an AUC of 0.897. The calibration curve demonstrated a high consistency between the predicted probability of CVD occurrence by the nomogram and the actual incidence. DCA further validated that the nomogram achieved a net clinical benefit at different threshold probabilities. Conclusion The radiomics nomogram established in this study demonstrates excellent performance in predicting the risk of DKD patients developing CVD. It holds promise to aid clinicians in achieving individualized management and early intervention for DKD patients.

Key words: Diabetic kidney disease, Cardiovascular disease, Shear wave elastography, Radiomics, Nomogram