Construction and Validation of a Prognostic Survival Prediction Model for NMIBC Patients Undergoing Radical Cystectomy

To screen predictors of prognosis in NMIBC patients treated with RC and construct a predictive model that accurately assesses their overall survival (OS). 1129 patients screened from the SEER database were randomized in a 7:3 ratio into a training group (790) and a validation group (339). Univariate and multivariate Cox regression analyses screened for prognostic factors, the best predictive model was determined by AIC minimum, and variables were examined for multicollinearity. Based on the total score obtained from the column line graph and the X-Tile procedure to find the best cutoff point and create a risk classification system. Finally, the model was evaluated and validated by C-index, AUC, drawing calibration curves (1000 bootstrap resamples), and Decision curve analysis (DCA). In the training group, the C-index was 0.67, and the ROC curves showed AUCs of 0.655, 0.704, and 0.722 for 1-, 5-, and 10-year OS, respectively; in the validation group, the AUCs were 0.738, 0.694, and 0.705, respectively. Meanwhile, the predictive performance of the present model was superior to that of the TNM staging, pLNR, and N staging, which can be used as a basis for patient counseling, follow-up scheduling, and treatment choice The model can provide a basis for patient counseling, follow-up scheduling and treatment selection.