To identify independent risk factors for urosepsis in diabetic patients with upper urinary tract stones (UUTS) and develop a prediction model to facilitate early detection and diagnosis, we retrospectively reviewed medical records of patients admitted between January 2020 and June 2023. Patients were divided based on the quick Sequential Organ Failure Assessment (qSOFA) score. The least absolute shrinkage and selection operator (LASSO) regression analysis was used for variable selection to form a preliminary model. The model was optimized and validated using the receiver operating characteristic (ROC) curve, the Hosmer-Lemeshow test and calibration curve, and decision curve analysis (DCA). A nomogram was constructed for visualization. A total of 434 patients were enrolled, with 66 cases and 368 controls. Six optimal predictors were identified: underweight, sarcopenia, poor performance status, midstream urine culture, urinary leukocyte count, and albumin-globulin ratio (AGR). The midstream urine culture was excluded due to its inability to provide rapid results. The final model demonstrated good prediction accuracy and clinical utility, with no significant difference in performance compared to the initial model. The study developed a prediction model for urosepsis risk in diabetic patients with UUTS, presenting a convenient tool for timely diagnosis, particularly in non-operated patients.