Construction and verification of delayed discharge prediction model for patients after ureteroscopic lithotripsy

TANG Ling; YUAN Yu

doi:10.12104/j.issn.1674-4748.2025.04.006

Abstract：Objective:To analyze the influencing factors of delayed discharge of patients after ureteroscopic lithotripsy,and to construct and verify the prediction model.Methods:A total of 240 patients who underwent ureteroscopic lithotripsis in hospital from January 2021 to January 2024 were retrospectively selected and divided into delayed group and no delayed group according to whether they were discharged late after surgery.The independent influencing factors of delayed discharge after ureteroscopic lithotripsy were screened by single factor and further Logistic regression analysis,and the prediction model was constructed and verified.Results:Logistic regression analysis showed that age,diabetes mellitus,operation time,catheter retention time,postoperative fever and postoperative pain were all independent influencing factors for delayed discharge of patients after ureteroscopic lithotripsy(all P<0.05).Based on this result,a risk prediction Nomogram model was constructed.The area under ROC curve of the modeling set was 0.948[95%CI(0.914,0.981)],the Yoden index was 0.802,the sensitivity and specificity were 0.914 and 0.888 respectively,the MAE in the calibration curve was 0.014,and the H-L goodness of fit test results showed χ2=4.808,P=0.778.The area under the ROC curve of the verification set was 0.973[95%CI(0.945,1.000)],the Yoden index was 0.886,the sensitivity and specificity were 1.000 and 0.886,respectively,and the MAE in the calibration curve was 0.023,indicating a good consistency of the model.Conclusion:Advanced age,diabetes mellitus,operation time and catheter retention time,and postoperative fever and pain are risk factors for delayed discharge after ureteroscopic lithotripsy.The predictive Nomogram model established by this method has high clinical application value.