Abstract: Objective To construct a visualization model for predicting the postoperative recurence of bladder cancer based on preoperative serum interleukin-6 ( IL-6)，prostaglandin E2 ( PGE2)，and tumor necrosis factor-alpha ( TNF-a) levels. Methods The cdlinical data of 348 patients with bladder cancer admitted to our hospital from June 2018 to February 2023 were retrospectively collected，and they were randomly divided into the training set (n= 232) and validation set (n= 116) ata ratio of 2:1 by the computer-generated random number table. All patients were followed up. The patients who had recurrence were included in the recurrence group, and those who did not have recurrence were included in the non-recurrence group. The levels of serum IL-6, PGE2, and TNF-a and general data of the recurrence group and non-recurrence group in the training set were compared. The Logistic regression model was used to analyze the infuencing factors of postoperative ecurrence of bladder cancer in the training set and est ablish a regression equation. The receiver operating characteristic ( ROC) curve was used to analyze the efcacy of preoperative IL-6, PGE2, and TNF-a alone and their combination in predicting the postoperative recurrence of bladder cancer. A risk prediction nomogram model for the postoperative recurence of bladder cancer was established and validated. Results Compared with the non-recurrence group, the levels of serum IL-6, PGE2 ,and TNF-a, tumor diameter, and proportions of multifocal tumors, tumor stage T2 ~Ts，and WH0 tumor pathological grade I~II in the recurrence group were increased, while the proportion of regular bladder perfusion after surgery was decreased ( P<0.05). Logistie regression analysis showed that preoperative serum IL-6, PGE2, and TNF-ox levels, tumor stage, and WHO tumor pathological grade were the infuencing factors of postoperative recurrence of bladder cancer ( P<0.05). The established Logistie regression equation was:Y= 1.718X +2.081X2+1.815X, +2.319X。+1.868X. The ROC curve showed that the optimal cut-off points of preoperative serum IL-6,PGE2, and TNF-a levels for predicting the postoperative recurrence of bladder cancer were 0.60 ng/L, 57.13 pg/mL, and 2. 10ng/ mL, respectively. The areas under the ROC curve ( AUCMR ) of senum IL-6, PGE2, and TNF-a alone and their combination were 0.729, 0.743, 0.733, and 0.825, respectively. Based on the structure of Logistice regression analysis in the training set, a risk nomogram model for predicting the postoperative recurence of bladder cancer was established. The prediction sensitivity, specificity, and AUCROC of the model in the training set and validation set were 94.12% and 90.20%，90.06% and 87.29%, and 0.940 and 0.914, respectively. The internal validation results of the Bootstrap method showed that the C-index values of the training set and validation set were 0.918 (95%CI:0.824-0.987) and 0.901 (95% CI:0.835-0.957), respectively. Conclusion Preoperative serum IL-6, PGE2 , and TNF-a levels are the influencing factors of postoperative recurence of bladder cancer, and the isk nomogram model based on them has good prediction efficacy.