Prediction of survival and prognostic factors in patients with bladder cancer after surgery using artificial intelligence recommendation algorithm: a preliminary study

Prediction of survival and prognostic factors in patients with bladder cancer after surgery using artificial intelligence recommendation algorithm: a preliminary study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Prediction of survival and prognostic factors in patients with bladder cancer after surgery using artificial intelligence recommendation algorithm: a preliminary study Yue Zhang, Ying Ke, Bo Yang, Xiang Gao, Lijie Wen, Ce Zhang, Yang Yu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7937665/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objective To discover the variables that affect bladder cancer (BC) patients' survival and prognosis after surgical treatment, and to use this knowledge to build an artificial intelligence (AI)-based recommendation algorithm. Methods This study comprised 832 BC patients who underwent surgery at The Second Affiliated Hospital of Dalian Medical University (2nd HDMU) and Nanfang Hospital of Southern Medical University (NHSMU) between January 2007 and January 2019. Their clinical and follow-up data were obtained. The 2nd HDMU patients were the training group, whereas NHSMU patients were the test group for external validation. An AI algorithm model was created using the deep neural network (DNN). The parameters influencing patient survival were analyzed and ranked with the assistance of AI algorithm. Results Out of the 832 bladder cancer patients included in this study, 438 (52.64%) were treated in the 2nd HDMU, while 394 (47.36%) were in the NHSMU. Among the BC cases, 579 (69.6%) were diagnostic of non-muscle invasive bladder cancer, while only 253 (30%) were muscle-invasive bladder cancer. In terms of surgical intervention, 539 (64.8%) patients underwent transurethral resection of bladder tumor, 66 (7.9%) received partial cystectomy, and 227 (27.3%) received total cystectomy. We concluded that the factors affecting the survival and prognosis of patients, in descending order, were T stage, pathological grade, hypertension or cardiovascular and cerebrovascular diseases, hemoglobin concentration, serum calcium, smoking, serum albumin level, lymphocyte count, age, serum albumin/globulin ratio, surgical method, N stage, and creatinine clearance rate. The testing group evaluated and confirmed this model to predict BC patients' survival before surgery. Conclusion Utilizing DNN modeling and external validation, the influencing factors of postoperative survival can be predicted for patients with BC. It can be employed to forecast BC patients' surgical outcomes before surgery. Additionally, this model can provide algorithmic assistance in selecting surgical and postoperative follow-up strategies for such patients. Bladder cancer Deep learning algorithm Surgery prognosis system Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 1 Introduction Bladder urothelial carcinoma (BUC) is the second most common genitourinary system cancer [ 1 , 2 ] . Transurethral resection of bladder tumor (TURBT) and total cystectomy are standard treatments for non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC), respectively [ 3 ] . In some circumstances, selective partial cystectomy based on specific parameters such as tumor size and site of tumor is indicated [ 4 ] . In recent years, some studies have found that neoadjuvant chemotherapy or immunotherapy, or chemoimmunotherapy therapy can effectively reduce tumor stage, allowing patients with muscle-invasive bladder cancer to undergo bladder preservation therapy and achieve a good prognosis [ 5 ] . Recent research suggests that the postoperative pathological stage and grade of the tumor can be employed as the primary survival prognostic variables for BUC patients, with age, gender, smoking history, etc. serving as secondary survival prognostic factors [ 6 , 7 ] . In recent years, studies have reported new preoperative prognostic markers such as neutrophil/lymphocyte ratio and serum albumin/globulin ratio, which also have certain values [ 6 , 8 ] . Unfortunately, few studies fully examine each prognostic factor and quantify the survival prognosis of patients with varied tumor burdens after undergoing different surgical regimens. In this study, we aim to develop a surgical prognosis prediction system to assess the survival and prognosis of BUC patients treated with various surgical techniques. Simultaneously, we will investigate the transformation path of artificial intelligence model software in order to implement the introduction of known factors prior to surgery and obtain the prediction output of survival prognosis of BUC patients undergoing different surgical procedures after model processing. Compared to conventional research, this approach allows patients to intuitively experience the therapeutic effects of various surgeries, enabling doctors and patients to jointly optimize treatment decisions, improve the efficacy of preoperative communication, and facilitate the development of individualized follow-up plans for patients. 2 Materials and methods 2.1 Study object and grouping We accessed the electronic medical record system of The Second Affiliated Hospital of Dalian Medical University (2nd HDMU) and Nanfang Hospital of Southern Medical University (NHSMU), and obtained the clinical information of 832 adult patients with proven BUC from January 2007 to January 2019. Inclusion criteria: Initial patients underwent surgical treatment. Postoperative pathological diagnosis was bladder urothelial carcinoma. Complete clinical and follow-up information was available. Postoperative follow-up for at least 3 years. Patients with retained bladders received a full course of epirubicin-based bladder transfusion chemotherapy after surgery. Exclusion criteria: Perioperative period underwent neoadjuvant or adjuvant radiation or chemotherapy. Clinical or follow-up information was not accessible. Postoperative follow-up time less than 3 years. 438 patients fro m the 2nd HDMU were included in the training group for model training, and 394 patients fro m the NHSMU were included in the test group for model verification. A comprehensive evaluation was conducted to determine whether the model was overfitting, and an examination of the model's generalization ability wa s carried out. 2.2 Data collection For survival analyses, routine clinical physiological and laboratory indicators, including common tumor markers and pathological indicators, were added to make the model more realistic. General patient information, preoperative examination data, and postoperative pathology reports were obtained from inpatient medical records, while follow-up data were obtained from outpatient medical records and telephone interviews. The following indicators were included in our study: age, gender, smoking, history of hypertension or cardiovascular and cerebrovascular diseases, history of diabetes, and surgical method. The pathological indicators include: tumor number, TNM (Tumor-Node-Metastasis classification) stage and pathological grade. Preoperative laboratory indicators include: white blood cell count, hemoglobin concentration, neutrophil count, lymphocyte count, neutrophil/lymphocyte ratio, platelet count, serum albumin level, serum globulin level, serum albumin/globulin ratio, alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, alkaline phosphatase, serum calcium, serum phosphorus and serum creatinine levels. Finally, creatinine clearance was calculated by the Cockcroft-Gault formula [ 9 ] and included as variable in this study. 2.3 Statistical analysis Statistical analysis was performed using SPSS (Statistical Package for the Social Sciences) 18.0 software. The quantitative data were represented by means and standard deviations. Chi-square test and two independent sample t-tests were used for single-factor screening. Variables with significant differences in univariate analysis were determined as input variables, and the 3-year overall survival (OS) of patients was determined as the output variables. The inspection level was set at 0.05. 2.4 Artificial intelligence learning model We use Python 3.7.4 to develop an artificial intelligence learning model for a deep neural network (DNN) which consists of an input layer, an output layer, and a hidden layer. In a large data set, the self-learning properties of the hidden layer are exploited to constantly improve classification accuracy by updating information and the network. Figure 1 depicts th e network topology, illustrating a dynamic relationship established between input features and output features to enable input-to-output prediction. Specifically, the purpose of this study is to establish the model relationship between the preoperative influencing factors, the choice of surgical method, and the patient's 3-year OS, with the aim of predicting and evaluating long-term patient survival. Additionally, this study aims to assign weights and ranks to each influencing factor based on it s importance and develop a versatile forecasting model. The creation of the core activation function is essential for the functioning of the DNN. In this example, a four-layer neural network is utilized. The input layer and hidden layer activation functions are Relu, whereas the output layer activation functions are Sigmoid. The training data set is utilized to establish the model, while the test data set is used to further evaluate the model. The accuracy and generalization capability of the model can prevent the model from being over-generalized due to the output being too sparse. Simultaneously, the Tensorflow module of the python2.7 software is employed, and the stochastic gradient descent algorithm is used to determine the weight of the influencing factors. In addition, we improve the model fitting effect through backpropagation, i.e., through the backpropagation of the error, compare and analyze the error between the result obtained by each training and the expected result, and iteratively optimize to find the minimum value using the gradient descent method. Relu activation function: $$\:f\left(x\right)=\text{m}\text{a}\text{x}(0,x)$$ Sigmoid activation function: $$\:f\left(z\right)=\frac{1}{1+{e}^{-z}}$$ Influencing factor weight calculation formula (stochastic gradient descent algorithm): $$\:{\omega\:}_{ji}\left(n\right)={\omega\:}_{ji}\left(n-1\right)-\eta\:{\Delta\:}{\omega\:}_{ji}\left(n-1\right)$$ Accuracy evaluation function: $$\:Accuracy=\frac{\text{T}\text{P}+\text{T}\text{N}}{\text{T}\text{P}+\text{T}\text{N}+\text{F}\text{P}+\text{F}\text{N}}$$ Notes: Actual category Prediction Category Yes No total Yes TP FN P (Actual Yes) No FP TN N (Actual No) total P’ (Classified as Yes) N’ (Classified as No) P + N TP: true positive (positive, prediction is also positive, prediction is correct) TN: true negative (negative, prediction is also negative, prediction is correct) FP: false positive (false positive, actually negative, predicted to be positive, predicted incorrectly) FN: false negative (false negative, actually positive, negative at prediction, wrong prediction) Notes : \(\:f\left(x\right)\) represents the output of the ReLU activation function, and x represents the input value. The principle of the ReLU function is to output 0 when the input is less than or equal to 0, and equal to the input value when the input is greater than 0. Among them, \(\:f\left(z\right)\) represents the output of the Sigmoid activation function, x represents the input value, and e is the base of the natural logarithm. The Sigmoid function effectively maps input values to a range between 0 and 1, making it particularly suitable for models whose outputs need to be interpreted as probabilities. Among them, w (n) represents the weight vector at the nth iteration. \(\:{\eta\:}\) is the learning rate used to control the step size of weight updates, and \(\:{\Delta\:}{{\omega\:}}_{\text{j}\text{i}}\left(\text{n}-1\right)\) represents the gradient of the loss function with respect to the weight vector w (n-1). The model parameter configuration utilized a four-layer network with two hidden layers (Fig. 2 ). The number of neurons in each layer was 64, 32, 16, and 1 respectively, and each layer adopted Dense full connection. The objective function was stochastic gradient descent, and 50 iterations were performed (Table 1 ). Data pre-processing (missing values and outliers) was performed using Python's Pandas module and Numpy module. Data exceeding 1.5 to 3 times the upper and lower quartile range were set as outliers. Data that exceeds the upper and lower quartile range by 3 times is set as extremum. The data quality of this study is high, with only a small number of extremum and outliers. Therefore, in this study, we removed the extreme values. As for the outliers, thos e greater than 1.5 times of the upper quartile was changed to 1.5 times of the upper quartile, and thos e less than 1.5 times of the lower quartile was adjuste d to 1.5 times of the lower quartile. For a small number of missing values, based on the normal distribution test results of the data, the normal distribution data was filled with the mean, while non normal distribution data was filled with the median. Table 1 Model Training Profile Model: “sequential” Layer(type) OutputShape Para# Dense_0(Dense) (None,64) 1088 Dense_1(Dense) (None,32) 2080 Dense_2(Dense) (None,16) 528 Dense_3(Dense) (None,1) 17 Total params:3713 Trainable params:3713 Non-trainable params:0 Notes: This example was a 4-layer neural network with 50 training iterations. 2.5 Contrast and verification of models To validate the model, we used 2nd HDMU data as the training set and NSMU data as the test set to establish Logistic regression, classification and regression tree (CART), and Bayesian modeling. And compare the performance parameters between the models. 3 Results 3.1 Basic information of patients Among the 832 patients, 438 (52.64%) were treated at the 2nd HDMU, and 394 (47.36%) were treated at NHSMU; the youngest patient was 19 years old, the oldest was 96 years old, and the average age was 64 years old. In term s of gender, 687 (82.6%) were males, while 145 (17.4%) were females. There were 482 (57.9%) solitary tumors and 350 (42.1%) multiple tumors. As for tumor pathology, there were 579 cases (69.6%) at Ta-T1 stage, 158 cases (19%) at T2 stage, 69 cases (8.3%) at T3 stage, and 26 cases (3.1%) at T4 stage. Among whole participants, there were 99 cases (11.9%) of low malignant potential, 284 cases of low grade (34.1%), and 449 cases of high grade (54%). In the study of treatment modalities, 539 (64.8%) cases received TURBT, 66 (7.9%) cases received partial cystectomy, and 227 (27.3%) cases received total cystectomy. The detailed comparison of the basic characteristics of the patient population in the two hospitals is shown in Table 2 . Apart from gender, tumor number, and N stage, there were significant differences in other characteristics between the two hospitals, but the trend was consistent. The differences were considered to be influenced by regional economy, population culture, as well as regional radiation effects between the hospitals. Table 2 Comparison of basic characteristics of patients undergoing bladder cancer surgery in two hospitals Influencing factors 2nd HDMU(n = 438) NHSMU(n = 394) P age Years old 66.63 ± 11.604 59.27 ± 14.291 <0.001 gender Female (%) 83(18.9) 62(15.7) Male (%) 355(81.1) 332(84.3) 0.222 Tumor multiplicity Unifocal (%) 244(55.7) 238(60.4) Multifocal (%) 194(44.3) 156(39.6) 0.17 Pathological Grade LMP (%) 4(0.9) 95(24.1) Low (%) 158(36.1) 126(32) High (%) 276(63) 173(43.9) <0.001 Pathological T stage Ta-T1(%) 329(75.1) 250(63.5) T2(%) 65(14.8) 93(23.6) T3(%) 32(7.3) 37(9.4) T4(%) 12(2.7) 14(3.5) 0.003 Pathological N stage N0(%) 428(97.7) 389(98.7) N1(%) 6(1.4) 2(0.5) N2(%) 4(0.9) 3(0.8) 0.431 Surgical method TURBT (%) 304(69.4) 235(59.6) partial cystectomy 18(4.1) 48(12.2) radical cystectomy 116(26.5) 111(28.2) <0.001 3-year OS Survival (%) 385(87.9) 364(92.4) Death (%) 53(12.1) 30(7.6) 0.031 LMP: low malignant potential, TURBT: transurethral bladder tumor resect, T: Tumor, N: Lymph node staging., OS: Overall survival. 3.2 Extraction of data characteristics The training group included 438 patients treated at the 2nd HDMU, and the data feature selection technique employed was univariate analysis. Patients were classified into survival and death groups based on 36-month survival after surgery. In univariate analysis, age, smoking, lymphocyte count, hemoglobin, albumin/globulin ratio, creatinine clearance rate, blood calcium, T stage, N stage, pathological grade, and surgical procedure differed significantly across groups (P < 0.05, Supplementary Table 1). These variables were fed into the DNN model. In addition, we included a history of hypertension or cardiovascular and cerebrovascular illnesses with a P-value of 0.079 (Table 3 ), which, while not statistically significant, was included due to its potential clinical significance. At the same time, we conducted collinearity diagnosis on the included variables. The collinearity diagnosis results showed that the variance inflation factors of each variable were all less than 10, with a maximum of 2.861 and a minimum of 1.068, indicating no collinearity between the independent variables. Table 3 Comparison of general information between the death group and the survival group of BUC patients in the 2nd HDMU 3 years after surgery Influencing factors death group(n = 53) survival group(n = 385) P age Years old 79.45 ± 10.079 64.85 ± 10.66 <0.001 gender Female (%) 12(22.6) 71(18.4) Male (%) 41(77.4) 314(81.6) 0.464 Hypertension/ Cardio-cerebrovascular Disease None (%) 28(52.8) 257(66.8) Yes (%) 25(47.2) 128(33.2) 0.079 diabetes None (%) 45(84.9) 341(88.6) Yes (%) 8(15.1) 44(11.4) 0.357 smoke None (%) 40(75.5) 233(60.5) Yes (%) 13(24.5) 152(39.5) 0.039 WBC g/L 6.47 ± 1.188 6.4 ± 1.805 0.717 NE g/L 4.11 ± 1.107 3.91 ± 1.514 0.341 LYM g/L 1.74 ± 0.539 1.95 ± 0.673 0.035 NE/LYM 2.77 ± 2.003 2.39 ± 2.412 0.273 HGB g/L 130.9 ± 17.828 141.15 ± 18.782 <0.001 PLT g/L 222.01 ± 70.329 216.26 ± 63.659 0.543 ALB g/L 38.12 ± 4.313 42.91 ± 4.249 <0.001 ALB/GLO 1.44 ± 0.357 1.66 ± 0.367 <0.001 ALT U/L 23 ± 13.368 22.71 ± 12.498 0.874 GOT U/L 21.73 ± 20.926 23.36 ± 16.709 0.520 LDH U/L 198.66 ± 39.742 198.66 ± 37.189 0.253 ALP U/L 75.67 ± 26.263 71.92 ± 29.553 0.381 CCr ml/(min*1.73m 2 ) 71.99 ± 19.398 81.98 ± 14.196 0.001 Serum Calcium mmol/L 2.18 ± 0.03 2.29 ± 0.01 <0.001 Phosphorus mmol/L 1.09 ± 0.249 1.13 ± 0.433 0.525 Tumor multiplicity Unifocal (%) 25(47.2) 219(56.9) Multifocal (%) 28(52.8) 166(43.1) 0.246 Pathological Grade LMP (%) 0 4(1.0) Low (%) 5(9.4) 153(39.7) High (%) 48(90.6) 228(59.2) <0.001 Pathological T stage Ta-T1(%) 22(41.5) 307(79.7) T2(%) 16(30.2) 49(12.7) T3(%) 9(17.0) 23(6.0) T4(%) 6(11.3) 6(1.6) <0.001 Pathological N stage N0(%) 47(88.7) 381(99.0) N1(%) 4(7.5) 2(0.5) N2(%) 2(3.8) 2(0.5) <0.001 Surgical method TURBT (%) 27(50.9) 277(71.9) partial cystectomy 10(18.9) 8(2.1) radical cystectomy 16(30.2) 100(26.0) <0.001 WBC: White blood cell count, NE: neutrophilic granulocyte, LYM: lymphocyte, HGB: hemoglobin, PLT: platelet, ALB: albumin, GLO: globular proteins, ALT: glutamic-pyruvic transaminase, GOT: glutamic oxaloacetic transaminase, LDH: lactate dehydrogenase, ALP: alkaline phosphatase, CCr: Creatinine Clearance Rate, LMP: low malignant potential, TURBT: transurethral bladder tumor resect, T: Tumor staging, N: Lymph node staging 3.3 DNN model establishment In the DNN modeling process, TensorFlow Keras framework was used, and the data result visualization used the TensorBoard modeling data visualization panel. Mapping the classification into three dimensions, Fig. 3 shows the modeling effect, effectively distinguishing whether or not the patients died at the follow-up cut-off point. The convergence of each neural network layer during model training is shown in Fig. 4 , indicating that the predicted value of the function closely fits the actual situation of the patients, with minima l error and prediction bias, thereby demonstrating the strong predictive efficacy of our model. The stochastic gradient descent algorithm was used to determine the relative importance of influencing factors. The results indicate that T stage has the highest influence on patient survival, followed by pathological grade, hypertension or cardiovascular and cerebrovascular diseases, hemoglobin concentration, serum calcium, smoking, serum albumin level, lymphocyte count, age, serum albumin/globulin ratio, surgical method, N stage, and creatinine clearance rate (Table 4 ). After establishing the DNN model, we randomly select 88 cases of data from 20% of the training set as internal validation. And the accuracy of the model was validated using data from NHSMU patients with bladder cancer. The areas under the ROC curves of the models are 0.890 (95% CI 0.828–0.951) for the training group (Fig. 5 ), 0.863 (95% CI 0.741–0.985) for the internal validation group (Fig. 6 ), and 0.835 (95% CI 0.744–0.926) for the external validation group (Fig. 7 ). The relevant ROC curves are shown in the figure. The fact that the accuracy rate of the model in the training group was 88.57% (95% CI: 0.8805–0.8909), the accuracy rate of the model in the verification group was 92.05% (95% CI: 0.9168–0.9242), and the accuracy rate for the entir e sample population was 90.56% (95% CI: 0.9017–0.9095). These results demonstrate that the DNN model has a high level of predictive performance in this instance. Additionally, a calibration curve of the DNN model was generated using Python (Fig. 8 ), which illustrates a good fit between the predicted patient survival and the actual condition. Table 4 The weight value of factors influencing postoperative survival and rank by importance factors importance T 0.5876238 pathological grading 0.5786078 Hypertension/ Cardio-cerebrovascular Disease 0.5035198 HGB 0.4522362 Serum Calcium 0.4269765 smoke 0.3287318 ALB 0.3209476 LYM 0.2988171 age 0.2210308 ALB/GLO 0.2157595 Surgical method 0.1922349 N 0.1830217 CCr 0.151636 3.4 Contrast and verification of models We conducted traditional predictive model modeling using the same set of data to verify the predictive performance of the DNN model. Firstly, we applied the 2nd HDMU data for Logistic multivariate analysis based on the influencing factors selected through univariate analysis, and the results are shown in supplementary table 3. Then Logistic regression, classification and regression tree (CART), and Bayesian modeling were performed separately. And the data of NHSMU was applied for validation. We compared and analyzed the accuracy, precision, sensitivity, and specificity of these models. (Table 6 ) It can be found that the DNN model was significantly superior to the other three methods in terms of accuracy, sensitivity and F score. And the models showed similar specificity. The receiver-operating-characteristic (ROC) curve of the models was drawn in Fig. 9 , and the area under curve (AUC) of the DNN model was 0.939, which was superior to the others. Table 5 Logistic multivariate analysis based on the influencing factors selected through univariate analysis in the 2nd HDMU Influencing factors OR 95% C.I.for EXP(B) P Lower Upper age 1.078 1.040 1.116 <0.001 Hypertension/ Cardio-cerebrovascular Disease 1.550 0.811 2.963 0.185 smoke 1.154 0.580 2.298 0.683 LYM 0.923 0.565 1.508 0.750 HGB 0.997 0.977 1.017 0.762 ALB 1.041 0.939 1.154 0.444 ALB/GLO 0.805 0.354 1.831 0.605 CCr 0.996 0.977 1.016 0.696 Serum Calcium 0.241 0.015 3.910 0.317 Pathological Grade 1.214 0.610 2.419 0.581 Pathological T stage 1.618 0.979 2.674 0.061 Pathological N stage 2.691 1.258 5.757 0.011 Surgical method 0.656 0.408 1.055 0.082 Table 6 Comparison of the Logistic regression model, the classification and regression tree model, the Bayesian model, and the DNN model Model Accuracy Precision Sensitivity Specificity F Score DNN 0.91 0.61 0.90 0.94 0.73 Logistic regression 0.82 0.63 0.17 0.97 0.26 Classification and regression tree 0.84 0.92 0.14 0.99 0.25 Bayesian 0.81 0.51 0.20 0.95 0.29 4 Discussion BUC is a common cancer with a high recurrence rate, and tumor pathological stage and grade are the most independent prognostic predictors for survival [ 2 ] . Recent years, through basic experiments and bioinformation analysis, some glycoproteins in urine, gene mutation sites in genetic testing, and prognostic markers based on transcriptome files have been found to be associate d with the proliferation, invasion, and metastasis of tumor cells, and can predict the clinical efficacy of treatment and the survival of patients. These indicators have accurate predictions, stable efficacy, and strong individual specificity. However, due to the limitations in detection technology and cost, they have not been popularized in clinical practice. Additionally, the prognostic value of clinical routine indicators for patients' conditions has been explored. For instance, serum albumin, serum albumin/globulin ratio, granulocyte/lymphocyte ratio have been identified as indicators that can be widely used in clinic without imposing additional economic burdens on patients. However, most of the current reports on survival prognosis are single-center, small-sample studies that used Logistic or Cox univariate/multivariate regression models for analysis. The conclusions obtained were inconsistent. Logistic and Cox models are generalized linear models. While few studies used non-linear artificial intelligence analysis to conduct interaction analysis of influencing factors and weight quantification. Theoretically, non-linear models have better predictive performance than linear models. Therefore, applying non-linear artificial intelligence analysis can comprehensively evaluate various prognostic indicators and establish a prognostic model with high accuracy and good universality. Such a model can be used to quantitatively evaluate the survival prognosis of patients with different tumor burdens after receiving various surgical treatments, which would be highly desirable. Neural network technology, originated in the 1950s, has an input layer, an output layer, and a hidden layer. Through the calculation of the hidden layer, each variable receives a weight in the hidden layer, and the final change result is sent to the output layer to generate the prediction result. Nowadays, several researchers are attempting to apply artificial intelligence technologies for disease diagnosis and therapy modeling. In the field of urological tumors, artificial intelligence technology has made many advances in improving the diagnosis and treatment of prostate cancer, kidney cancer, and bladder cancer. Treatment plan optimization and patient follow-up education are not well-researched and the technology is still in its infancy [ 10 ] . This study uses the DNN algorithm to model, mainly relying on the method of backpropagation to improve model fitting. Through repeated training, the results obtained each time are compared with the expected results for error analysis. Subsequently, the weight and threshold of each neural node are adjusted based on the comparison results, enabling the model to progressively approach the expected result and improve its accuracy. In the end, the test accuracy rate of the model training group in this example was 88.57%, and the test accuracy rate of the external verification group was 92.05%, and the calibration curve shown that the patient survival predicted by the DNN model has a good fit to the actual condition, which proves that the model has good predictive performance, and subsequent software development can be performed according to the weight. We also used the same data for Logistic regression, CART, and Bayesian modeling separately. Through comparing and analyzing the DNN model with the other models, we’ve found that the DNN model is significantly superior to other three methods in F score, and ha s a better AUC than others. This indicates that the DNN model in this study has better prediction performance for the long-term survival of BUC patients after surgery and is suitable for clinical reference. With the exception of gender, tumor number, and N stage, there were significant differences in the other characteristics of the bladder cancer patients in the two hospitals in this study. The reasons were considered to be the economic and cultural level of the population between regions, as well as the influence of hospitals. That is to say, people in the developed area (Guangzhou) pay more attention to their own physical condition, undergo regular physical examination and seek medical treatment at the early stage of the disease (such as the first occurrence of hematuria). As a result, the average age of the patient population is younger, and the pathological grade is lower. At the same time, patients' higher economic level and higher requirements for quality of life promote the preference for more influential provincial hospitals. Therefore, the T stage of bladder cancer patients in NHSMU is higher, and the proportion of partial cystectomy and total cystectomy is higher than that in the 2nd HDMU. Although there are statistical differences in the basic characteristics of patients between regions, the trend of the characteristics is consistent and the difference is not large. The data of patients from the north and the south were used for modeling and verification, which can verify the generality of the model obtained for Chinese bladder cancer patients. In this work, the 3-year OS was utilized as the research endpoint for univariate analysis, and the indicators with statistically significant differences between survival and death patients were incorporated into the DNN model for analysis. Age is the fundamental metric that indicates physical reserve, and increased age is a risk factor for the short- and long-term survival of nearly all malignant tumor diseases [ 11 ] . The history of hypertension or cardiovascular and cerebrovascular illnesses reveals the patient's underlying illness, which can considerably impact the patient's risk of mortality from all causes [ 12 ] . Smoking also is a clear risk factor for bladder cancer recurrence [ 13 ] . The current study provides evidence that vitamin D deficiency may be related to tumor recurrence and progression, and blood calcium level can indirectly reflect the vitamin D level in patients; therefore, low calcium may be a risk factor affecting the survival of cancer patients [ 14 , 15 ] . Tumor pathological T stage, N stage, and grading reflect the tumor burden and malignancy, the higher the stage and grading, the worse the prognosis [ 16 ] . The level of hemoglobin, serum albumin, and albumin/globulin ratio can reflect the nutritional level of the patient and the degree of tumor consumption. Malnutrition is considered to be the main reason for the high incidence of postoperative complications, which can lead to the weakening of the patient's defense mechanism. Multiple studies have confirmed that anemia and hypoalbuminemia are adverse prognostic factors for patients requiring surgery. Complications, and rapid cancer progression are more common in these patients [ 6 , 17 , 18 ] . Lymphocyte count reflects the immune ability of the body, and studies have found that the decline of the body's immune ability is related to the poor prognosis of tumors [ 18 ] . The creatinine clearance rate reflects the patient's renal function and the level of basic physical quality, making it one of the predictors of survival prognosis. The above factors were confirmed to be related to the survival and prognosis of BUC patients through univariate analysis and were incorporated into the DNN model for further analysis. In this case, the DNN model analysis found that the top three factors affecting the patient's surgical prognosis are: tumor T stage, pathological grade, and hypertension or cardiovascular and cerebrovascular diseases, which have a higher predictive effect on the long-term survival of patients than other indicators. The higher the pathological stage and tumor grade, the worse the underlying disease state, and the worse the prognosis of patients after surgery, which is consistent with the previous research results. Other influencing factors, in descending order of importance, were hemoglobin concentration, serum calcium, smoking, serum albumin level, lymphocyte count, age, serum albumin/globulin ratio, surgical method, N stage, and creatinine clearance rate. Different statistical methods may lead to differences in the calculation results of weights. It can be seen that the Logistic and DNN model diffe r somewhat in the weight ranking of the influencing factors. The influencing factor with the highest OR value obtained from Logistic analysis is N stage, followed by T stage, hypertension or cardiovascular and cerebrovascular diseases, and pathological grade, which is different from the DNN model. In clinical practice, T stage, N stage, and pathological grade are all important influencing factors that significantly affect the choice of surgical method and postoperative survival of patients. Regardless of the T stage and pathological grade of the patient, the presence of lymph node metastasis indicates poor prognosis. The survival time of patients in N1 or N2 stage is significantly shorter than N0 stage. However, only a few patients with advanced BUC have lymph node metastasis. In this study, only 27(3.2%) patients were classified as N1 or N2 in N stage, and the 3-year mortality rate was as high as 29.6% (8 cases). In clinical practice, clinicians pay more attention to the T stage and pathological grading of N0 BUC patients. Therefore, T stage and pathological grade may have greater prognostic value than N stage for the overall BUC patients. Therefore, we believe that the weight ranking of influencing factors using the DNN model is closer to clinical practice compared to the Logistic model. The number of follow-up cases in this study is large, and it is a multi-center study. The follow-up data of NHSMU were included for external verification. Through external verification, the DNN model which has an accuracy rate of 92% in predicting the 3-year OS of BUC patients, proves that the model in this study has good predictive performance. According to the corresponding weight of each prognostic factor, the software is being developed. By inputting the value of each influencing factor, the corresponding 3-year all-cause survival rate of the patient can be obtained, and the difference in the long-term survival rate after different surgical procedures can be calculated. This model has the characteristics of high accuracy and strong quantitative comparison, and it shows excellent application prospects in clinical work. 5 Limitations and future prospectives 1.This article is a preliminary exploration of the DNN artificial intelligence algorithm in the survival and prognosis of BUC surgery. The source of data is limited, and there may be differences in the baseline conditions of patients and the technical level of doctors. 2. All patients did not receive adjuvant or neoadjuvant therapy, but the adjuvant therapy plan can be included as one of the influencing factors for artificial intelligence learning to improve the survival prognosis model in further studies. In the future, the research team will use the self-learning nature of the DNN artificial intelligence model to incorporate patient data from more centers. They will enrich the prognostic factors, adjust the weight of the model neuron unit, improve the prediction accuracy of the model, and increase its survival prediction efficiency, thereb y providin g different surgical options for patients. The intuitive preoperative prediction of the treatment effect facilitates preoperative communication between doctors and patients, enabling them to select appropriate treatment methods and formulate individualized follow-up plans. 6 Conclusions Using DNN modeling and external validation, BUC patients' postoperative survival variables can be predicted. This can predict bladder cancer surgery outcomes before surgery. Additionally, this model can help patients choose surgery and postoperative follow-up techniques using algorithms. Declarations Author contributions Yang Yu is the guarantor. Yue Zhang, Ce Zhang, and Lijie Wen conceived, designed, and collected the questionnaire for the patient. Bo Yang and Xiang Gao collated data and produced graphs. Yue Zhang and Ying Ke drafted the article and embellished it with revisions. Yang Yu provided funding support. All authors made substantial contributions to the article draft and critically revised it. All authors approved the submitted and final versions. Funding This study was supported by the “1+X” program for Cross-disciplinary Innovation Projects of the Second Hospital of Dalian Medical University (2022JCXKZD05), In-Hospital Training Fund of the Second Hospital of Dalian Medical University (dy2yynpy202220), Liaoning Provincial Department of Education 2021 Scientific Research Funding Project (LJKZ0873), the Second Hospital of Dalian Medical University - Dalian Institute of Chemical Physics, Chinese Academy of Sciences "Collaborative Innovation Centre for Individualized Diagnosis and Treatment" Jointly Funded Project (UF-ZD-202014). Acknowledgments We thank all the participants who were willing to be followed up during this period. Ethics approval and consent to participate All procedures performed in this study were in accordance with the ethical standards of the Declaration of Helsinki. Approval was granted by the ethics committee of the Second Hospital of Dalian Medical University (Approval No. 177, 2023). Informed consent was obtained from all individual participants included in the study. Patient and Public Involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Conflict of Interest All authors have no competing interest to declare. data availability statement All data generated or analyzed during this study are included in this published article. References SIEGEL RL, MILLER K D, FUCHS H E et al. Cancer statistics, 2022[J/OL]. CA: A Cancer Journal for Clinicians, 2022, 72(1): 7–33. 10.3322/caac.21708 XIA C, DONG X, LI H, et al. Cancer statistics in China and United States, 2022: profiles, trends, and determinants[J/OL]. Chin Med J. 2022;135(5):584–90. 10.1097/CM9.0000000000002108 . GONTERO P. European Association of Urology Guidelines on Non–muscle-invasive Bladder Cancer (TaT1 and Carcinoma In Situ)—A Summary of the 2024 Guidelines Update[J]. invasive Bladder Cancer. ANTAR R M, XU V E, FARAG C M, et al. Oncologic outcomes of neoadjuvant chemotherapy and lymph node dissection with partial cystectomy for muscle-invasive bladder cancer[J/OL]. Translational Androl Urol. 2024;13(8). 10.21037/tau-24-165 . HU J, CHEN J, OU Z, et al. Neoadjuvant immunotherapy, chemotherapy, and combination therapy in muscle-invasive bladder cancer: A multi-center real-world retrospective study[J/OL]. Cell Rep Med. 2022;3(11):100785. 10.1016/j.xcrm.2022.100785 . YUE Z, FAN FEIL. Prognostic value of preoperative serum albumin in patients with non-muscle-invasive bladder cancer undergoing transurethral resection of bladder tumor [J/OL]. J South Med Univ. 2018;38(2):192. 10.3969/j.issn.1673-4254.2018.02.12 . Epidemiology LOBON. Screening, and Prevention of Bladder Cancer[J/OL]. 2022. 10.1016/j.euo.2022.10.003 OFNER H, LAUKHTINA E HASSLERMR, et al. Blood-Based Biomarkers as Prognostic Factors of Recurrent Disease after Radical Cystectomy: A Systematic Review and Meta-Analysis[J/OL]. Int J Mol Sci. 2023;24(6):5846. 10.3390/ijms24065846 . FERNANDEZ-PRADO R. Creatinine Clearance Is Not Equal to Glomerular Filtration Rate and Cockcroft-Gault Equation Is Not Equal to CKD-EPI Collaboration Equation[J/OL]. Am J Med. 2016;129(12). 10.1016/j.amjmed.2016.08.019 . WENHAO X, XI T, Aihetaimujiang A, et al. A systematic review of current advancements of artificial intelligence in genitourinary cancers [. J] China Oncol. 2022;32(1). 10.19401/j.cnki.1007-3639.2022.01.009 . CHANGFA X. Fractions and trends of cancer burden attributable to population ageing in China [J]. Chin J Oncol. 2022;44(1):79–85. BRAY F, LAVERSANNE M. The ever-increasing importance of cancer as a leading cause of premature death worldwide[J/OL]. Cancer. 2021;127(16):3029–30. 10.1002/cncr.33587 . LISHA L, HANG HANGHANGL. An analysis of disease burden of prostate, bladder and kidney cancers attributable to smoking in China from 1990 to 2019 [J]. Chin J Evidence-Based Med. 2022;22(05):530–6. 10.7507/1672-2531.202201059 . KEUM N, LEE D H, GREENWOOD D C, et al. Vitamin D supplementation and total cancer incidence and mortality: a meta-analysis of randomized controlled trials[J/OL]. Ann Oncol. 2019;30(5):733–43. 10.1093/annonc/mdz059 . Boot IWA, Wesselius A, Yu EYW, et al. Dietary vitamin D intake and the bladder cancer risk: A pooled analysis of prospective cohort studies. Clin Nutr. 2023;42(8):1462–74. 10.1016/j.clnu.2023.05.010 . VAN BRUWAENE S, COSTELLO A J, VAN POPPEL H. Prognosis of node-positive bladder cancer in 2016[J]. Minerva Urologica E Nefrologica = The. Italian J Urol Nephrol. 2016;68(2):125–37. GRIMM T, BUCHNER A, SCHNEEVOIGT B, et al. Impact of preoperative hemoglobin and CRP levels on cancer-specific survival in patients undergoing radical cystectomy for transitional cell carcinoma of the bladder: results of a single-center study[J/OL]. World J Urol. 2016;34(5):703–8. 10.1007/s00345-015-1680-7 . Zhang S, Du J, Zhong X, et al. The prognostic value of the systemic immune-inflammation index for patients with bladder cancer after radical cystectomy. Front Immunol. 2022;13:1072433. 10.3389/fimmu.2022.1072433 . Published 2022 Nov 29. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7937665","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":551222193,"identity":"92b7d96c-607e-4aca-803e-cc92a62b14c1","order_by":0,"name":"Yue Zhang","email":"","orcid":"","institution":"The Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yue","middleName":"","lastName":"Zhang","suffix":""},{"id":551222194,"identity":"cef5792f-32a0-4d84-81b7-b9486a24fa15","order_by":1,"name":"Ying Ke","email":"","orcid":"","institution":"The Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Ke","suffix":""},{"id":551222195,"identity":"54c720a4-a91e-4e4b-acef-63babf1476ef","order_by":2,"name":"Bo Yang","email":"","orcid":"","institution":"The Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Bo","middleName":"","lastName":"Yang","suffix":""},{"id":551222196,"identity":"c06c68af-7703-481d-9866-6dc9d0ed2485","order_by":3,"name":"Xiang Gao","email":"","orcid":"","institution":"The Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiang","middleName":"","lastName":"Gao","suffix":""},{"id":551222197,"identity":"cbe11124-081a-4750-bb41-4d670151e4c5","order_by":4,"name":"Lijie Wen","email":"","orcid":"","institution":"The Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lijie","middleName":"","lastName":"Wen","suffix":""},{"id":551222198,"identity":"3f8fc357-5eba-43ba-ae65-e4c3f5a1c4ac","order_by":5,"name":"Ce Zhang","email":"","orcid":"","institution":"Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ce","middleName":"","lastName":"Zhang","suffix":""},{"id":551222199,"identity":"406d260e-edc9-4505-901d-b12e5e507458","order_by":6,"name":"Yang Yu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIie3RIQvCQBTA8TcOzvLmjDcU/QonwpLgV7mVJQXjBYMDOYPaDX4Io3EgzPLsi9oFtdl0ZmWbzXC/fH/u3j0Ay/pDvHO9JTf9RK8WxyelJ+VJHVT3vCbW8pf7vTxRWp60QfV6rmF9mUWRf56xCg+DJGoicZREgQ6nHLz5QhUnzjT1NxrRPyyDLNy1QNBxW5wwx4gLCaznt2QhcZBiVJJwxoVrJEI2DMahYRUS5KzrGoWNfHyolgh08k9O8P3JQlGKpbN01gj5KpPBe5X3h560vfmqOPmAvx23LMuyvnoBfF9LHexmWv4AAAAASUVORK5CYII=","orcid":"","institution":"The Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Yang","middleName":"","lastName":"Yu","suffix":""}],"badges":[],"createdAt":"2025-10-24 07:08:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7937665/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7937665/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97142873,"identity":"c21c6213-2668-4560-ae4b-aa06366bbc43","added_by":"auto","created_at":"2025-12-01 10:08:02","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":350001,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/ea80cc3cc0b5ae32e8adc4a8.docx"},{"id":97124274,"identity":"ee3b9a09-d606-4181-b7c0-f1a024a02a85","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8828,"visible":true,"origin":"","legend":"","description":"","filename":"7ea476b38e8340bb99ff1029e7b948b8.json","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/54ceeb004436a6b6af91500f.json"},{"id":97124278,"identity":"741fb697-66dc-4a7f-ad4b-dc34363dbf1b","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":121733,"visible":true,"origin":"","legend":"","description":"","filename":"7ea476b38e8340bb99ff1029e7b948b81enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/6d7c93cd030aa1876a6db45a.xml"},{"id":97124291,"identity":"4f7f63ca-23b3-43d6-a54a-b4c27adc1857","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40061,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/a75bd90043373623573b9691.png"},{"id":97141959,"identity":"3a4be97d-4411-408a-b504-0892e606dda3","added_by":"auto","created_at":"2025-12-01 10:07:13","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12244,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/73d56330eedc046c9cbbcc73.png"},{"id":97142911,"identity":"390945d8-e317-4ecb-8073-c3c884321997","added_by":"auto","created_at":"2025-12-01 10:08:06","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5558,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/0b146514932469f39288e03b.png"},{"id":97124290,"identity":"82d8eabd-4507-4428-a1a8-521656667313","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16366,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/daf8be801fd2c5b4979a811c.png"},{"id":97124284,"identity":"2e910c83-5201-46dc-99b7-ca75f8f1ede8","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11475,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/f08cb3a78bc80d8757302840.png"},{"id":97124297,"identity":"01eaa81d-4d3c-4478-ab48-39e4eb628cc1","added_by":"auto","created_at":"2025-12-01 08:10:32","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12963,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/d8a10b3d6aed216e011a43e8.png"},{"id":97142972,"identity":"18a34273-59e5-40ea-a05f-cf1df12a91fc","added_by":"auto","created_at":"2025-12-01 10:08:10","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10628,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/155e77d8f46ea629d1deade2.png"},{"id":97124293,"identity":"6e37e2d7-c16c-4bf5-be04-4bef41fe53cd","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5373,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/9aac80b21b3b505a76b9bed9.png"},{"id":97124285,"identity":"1d4627a3-b91c-403a-8b00-64aaeec35da5","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20173,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/fe1186ecc1b14b6e2b9763b5.png"},{"id":97124296,"identity":"86452454-6e59-45a5-8fe1-23b5c2c6bbce","added_by":"auto","created_at":"2025-12-01 08:10:32","extension":"xml","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":119014,"visible":true,"origin":"","legend":"","description":"","filename":"7ea476b38e8340bb99ff1029e7b948b81structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/0fb074350e9e13acd4423ed0.xml"},{"id":97124294,"identity":"fc9d3545-0134-4ff6-940c-aff722563730","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"html","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":128926,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/c1b7f7a82d57266877c43e9c.html"},{"id":97141852,"identity":"47d0db53-273d-451a-8420-c0b39cf774dc","added_by":"auto","created_at":"2025-12-01 10:07:06","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":86193,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic diagram of the structure of a deep neural network\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/84c8efd4d23ed3ec062351f0.jpeg"},{"id":97124276,"identity":"feb55420-651e-40d5-95d1-c546780e65de","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":27158,"visible":true,"origin":"","legend":"\u003cp\u003eDeep learning neural network model architecture\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/253f53b46b4ffed45c513a6b.jpeg"},{"id":97249226,"identity":"cfe43718-2317-4f0d-9baa-2b8ff4f44ec4","added_by":"auto","created_at":"2025-12-02 13:11:30","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":19676,"visible":true,"origin":"","legend":"\u003cp\u003eThe spatial distribution of each data point in three perimeters\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/c233066d87bfb599a2ed4da6.jpeg"},{"id":97142727,"identity":"5bcf2cc4-6584-4a23-a259-d8ebbd6259f6","added_by":"auto","created_at":"2025-12-01 10:07:55","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":27816,"visible":true,"origin":"","legend":"\u003cp\u003eConvergence of each neural network layer during model training\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/f7640efdda85b407b749ce09.jpeg"},{"id":97141777,"identity":"955c9b72-1b39-4c44-a4c8-1be2d6ba2816","added_by":"auto","created_at":"2025-12-01 10:07:00","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":17814,"visible":true,"origin":"","legend":"\u003cp\u003eTraining group\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/f98e86bbe016c1de6fa78feb.jpeg"},{"id":97141714,"identity":"3d35156d-2367-4578-adbd-c85e2980f002","added_by":"auto","created_at":"2025-12-01 10:06:54","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":15326,"visible":true,"origin":"","legend":"\u003cp\u003eInternal validation\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/8c9fc69868f892924c13ad29.jpeg"},{"id":97124283,"identity":"1c633058-af8a-418f-83e7-d8e760a55b8f","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":16197,"visible":true,"origin":"","legend":"\u003cp\u003eExternal validation\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/1f34c50e83653b4277c3b7bf.jpeg"},{"id":97124287,"identity":"77048f72-dff9-4098-9a48-e143fba98828","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"jpeg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":15517,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curve of deep learning neural network model\u003c/p\u003e","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/4a6dcedee80891e2437baaed.jpeg"},{"id":97124292,"identity":"2729d08a-ff03-40e1-bfe4-a122b1fc62d2","added_by":"auto","created_at":"2025-12-01 08:10:31","extension":"jpeg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":38667,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver-operating-characteristic curve of the Logistic regression model, the classification and regression tree model, the Bayesian model, and the DNN model\u003c/p\u003e","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/7985b0436f0224bb37424af9.jpeg"},{"id":107704831,"identity":"e2833328-01f9-4551-96b0-a9c5a99e781b","added_by":"auto","created_at":"2026-04-24 08:59:52","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":804008,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7937665/v1/a431623d-a64c-4741-a17a-104a7d627399.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Prediction of survival and prognostic factors in patients with bladder cancer after surgery using artificial intelligence recommendation algorithm: a preliminary study","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eBladder urothelial carcinoma (BUC) is the second most common genitourinary system cancer\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. Transurethral resection of bladder tumor (TURBT) and total cystectomy are standard treatments for non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC), respectively\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. In some circumstances, selective partial cystectomy based on specific parameters such as tumor size and site of tumor is indicated\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e. In recent years, some studies have found that neoadjuvant chemotherapy or immunotherapy, or chemoimmunotherapy therapy can effectively reduce tumor stage, allowing patients with muscle-invasive bladder cancer to undergo bladder preservation therapy and achieve a good prognosis\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e. Recent research suggests that the postoperative pathological stage and grade of the tumor can be employed as the primary survival prognostic variables for BUC patients, with age, gender, smoking history, etc. serving as secondary survival prognostic factors\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. In recent years, studies have reported new preoperative prognostic markers such as neutrophil/lymphocyte ratio and serum albumin/globulin ratio, which also have certain values\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e. Unfortunately, few studies fully examine each prognostic factor and quantify the survival prognosis of patients with varied tumor burdens after undergoing different surgical regimens.\u003c/p\u003e\u003cp\u003eIn this study, we aim to develop a surgical prognosis prediction system to assess the survival and prognosis of BUC patients treated with various surgical techniques. Simultaneously, we will investigate the transformation path of artificial intelligence model software in order to implement the introduction of known factors prior to surgery and obtain the prediction output of survival prognosis of BUC patients undergoing different surgical procedures after model processing. Compared to conventional research, this approach allows patients to intuitively experience the therapeutic effects of various surgeries, enabling doctors and patients to jointly optimize treatment decisions, improve the efficacy of preoperative communication, and facilitate the development of individualized follow-up plans for patients.\u003c/p\u003e"},{"header":"2 Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Study object and grouping\u003c/h2\u003e\u003cp\u003e We accessed the electronic medical record system of The Second Affiliated Hospital of Dalian Medical University (2nd HDMU) and Nanfang Hospital of Southern Medical University (NHSMU), and obtained the clinical information of 832 adult patients with proven BUC from January 2007 to January 2019.\u003c/p\u003e\u003cp\u003eInclusion criteria:\u003c/p\u003e\u003cp\u003e\u003col style=\"list-style-type:lower-alpha;\"\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eInitial patients underwent surgical treatment.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePostoperative pathological diagnosis was bladder urothelial carcinoma.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eComplete clinical and follow-up information was available.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePostoperative follow-up for at least 3 years.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePatients with retained bladders received a full course of epirubicin-based bladder transfusion chemotherapy after surgery.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003eExclusion criteria:\u003c/p\u003e\u003cp\u003e\u003col style=\"list-style-type:lower-alpha;\"\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePerioperative period underwent neoadjuvant or adjuvant radiation or chemotherapy.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eClinical or follow-up information was not accessible.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003ePostoperative follow-up time less than 3 years.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003e438 patients fro\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003em\u003c/span\u003e the 2nd HDMU were included in the training group for model training, and 394 patients fro\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003em\u003c/span\u003e the NHSMU were included in the test group for model verification. A comprehensive evaluation was conducted to determine whether the model was overfitting, and an examination of the model's generalization ability wa\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003es\u003c/span\u003e carried out.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Data collection\u003c/h2\u003e\u003cp\u003eFor survival analyses, routine clinical physiological and laboratory indicators, including common tumor markers and pathological indicators, were added to make the model more realistic. General patient information, preoperative examination data, and postoperative pathology reports were obtained from inpatient medical records, while follow-up data were obtained from outpatient medical records and telephone interviews.\u003c/p\u003e\u003cp\u003eThe following indicators were included in our study: age, gender, smoking, history of hypertension or cardiovascular and cerebrovascular diseases, history of diabetes, and surgical method. The pathological indicators include: tumor number, TNM (Tumor-Node-Metastasis classification) stage and pathological grade. Preoperative laboratory indicators include: white blood cell count, hemoglobin concentration, neutrophil count, lymphocyte count, neutrophil/lymphocyte ratio, platelet count, serum albumin level, serum globulin level, serum albumin/globulin ratio, alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, alkaline phosphatase, serum calcium, serum phosphorus and serum creatinine levels. Finally, creatinine clearance was calculated by the Cockcroft-Gault formula\u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e and included as variable in this study.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Statistical analysis\u003c/h2\u003e\u003cp\u003eStatistical analysis was performed using SPSS (Statistical Package for the Social Sciences) 18.0 software. The quantitative data were represented by means and standard deviations. Chi-square test and two independent sample t-tests were used for single-factor screening. Variables with significant differences in univariate analysis were determined as input variables, and the 3-year overall survival (OS) of patients was determined as the output variables. The inspection level was set at 0.05.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Artificial intelligence learning model\u003c/h2\u003e\u003cp\u003eWe use Python 3.7.4 to develop an artificial intelligence learning model for a deep neural network (DNN) which consists of an input layer, an output layer, and a hidden layer. In a large data set, the self-learning properties of the hidden layer are exploited to constantly improve classification accuracy by updating information and the network. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e depicts th\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ee\u003c/span\u003e network topology, illustrating a dynamic relationship established between input features and output features to enable input-to-output prediction. Specifically, the purpose of this study is to establish the model relationship between the preoperative influencing factors, the choice of surgical method, and the patient's 3-year OS, with the aim of predicting and evaluating long-term patient survival. Additionally, this study aims to assign weights and ranks to each influencing factor based on it\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003es\u003c/span\u003e importance and develop a versatile forecasting model.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe creation of the core activation function is essential for the functioning of the DNN. In this example, a four-layer neural network is utilized. The input layer and hidden layer activation functions are Relu, whereas the output layer activation functions are Sigmoid. The training data set is utilized to establish the model, while the test data set is used to further evaluate the model. The accuracy and generalization capability of the model can prevent the model from being over-generalized due to the output being too sparse. Simultaneously, the Tensorflow module of the python2.7 software is employed, and the stochastic gradient descent algorithm is used to determine the weight of the influencing factors. In addition, we improve the model fitting effect through backpropagation, i.e., through the backpropagation of the error, compare and analyze the error between the result obtained by each training and the expected result, and iteratively optimize to find the minimum value using the gradient descent method.\u003c/p\u003e\u003cp\u003eRelu activation function:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:f\\left(x\\right)=\\text{m}\\text{a}\\text{x}(0,x)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eSigmoid activation function:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:f\\left(z\\right)=\\frac{1}{1+{e}^{-z}}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eInfluencing factor weight calculation formula (stochastic gradient descent algorithm):\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:{\\omega\\:}_{ji}\\left(n\\right)={\\omega\\:}_{ji}\\left(n-1\\right)-\\eta\\:{\\Delta\\:}{\\omega\\:}_{ji}\\left(n-1\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAccuracy evaluation function:\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:Accuracy=\\frac{\\text{T}\\text{P}+\\text{T}\\text{N}}{\\text{T}\\text{P}+\\text{T}\\text{N}+\\text{F}\\text{P}+\\text{F}\\text{N}}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eNotes:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"4\" rowspan=\"5\"\u003e\u003cp\u003eActual category\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c5\" namest=\"c2\"\u003e\u003cp\u003ePrediction Category\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003etotal\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eFN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP (Actual Yes)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eTN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eN (Actual No)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003etotal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eP\u0026rsquo; (Classified as Yes)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eN\u0026rsquo; (Classified as No)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP\u0026thinsp;+\u0026thinsp;N\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eTP: true positive (positive, prediction is also positive, prediction is correct)\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eTN: true negative (negative, prediction is also negative, prediction is correct)\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eFP: false positive (false positive, actually negative, predicted to be positive, predicted incorrectly)\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eFN: false negative (false negative, actually positive, negative at prediction, wrong prediction)\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cb\u003eNotes\u003c/b\u003e:\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:f\\left(x\\right)\\)\u003c/span\u003e\u003c/span\u003e represents the output of the ReLU activation function, and x represents the input value. The principle of the ReLU function is to output 0 when the input is less than or equal to 0, and equal to the input value when the input is greater than 0.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAmong them, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:f\\left(z\\right)\\)\u003c/span\u003e\u003c/span\u003e represents the output of the Sigmoid activation function, x represents the input value, and e is the base of the natural logarithm. The Sigmoid function effectively maps input values to a range between 0 and 1, making it particularly suitable for models whose outputs need to be interpreted as probabilities.\u003c/p\u003e\u003cp\u003eAmong them, w (n) represents the weight vector at the nth iteration. \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\eta\\:}\\)\u003c/span\u003e\u003c/span\u003e is the learning rate used to control the step size of weight updates, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}{{\\omega\\:}}_{\\text{j}\\text{i}}\\left(\\text{n}-1\\right)\\)\u003c/span\u003e\u003c/span\u003e represents the gradient of the loss function with respect to the weight vector w (n-1).\u003c/p\u003e\u003cp\u003eThe model parameter configuration utilized a four-layer network with two hidden layers (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The number of neurons in each layer was 64, 32, 16, and 1 respectively, and each layer adopted Dense full connection. The objective function was stochastic gradient descent, and 50 iterations were performed (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eData pre-processing (missing values and outliers) was performed using Python's Pandas module and Numpy module. Data exceeding 1.5 to 3 times the upper and lower quartile range were set as outliers. Data that exceeds the upper and lower quartile range by 3 times is set as extremum. The data quality of this study is high, with only a small number of extremum and outliers. Therefore, in this study, we removed the extreme values. As for the outliers, thos\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ee\u003c/span\u003e greater than 1.5 times of the upper quartile was changed to 1.5 times of the upper quartile, and thos\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ee\u003c/span\u003e less than 1.5 times of the lower quartile was adjuste\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ed\u003c/span\u003e to 1.5 times of the lower quartile. For a small number of missing values, based on the normal distribution test results of the data, the normal distribution data was filled with the mean, while non normal distribution data was filled with the median.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eModel Training Profile Model: \u0026ldquo;sequential\u0026rdquo;\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLayer(type)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOutputShape\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePara#\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDense_0(Dense)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(None,64)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1088\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDense_1(Dense)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(None,32)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2080\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDense_2(Dense)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(None,16)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e528\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDense_3(Dense)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(None,1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e17\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"3\"\u003eTotal params:3713\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"3\"\u003eTrainable params:3713\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"3\"\u003eNon-trainable params:0\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd colspan=\"3\"\u003eNotes: This example was a 4-layer neural network with 50 training iterations.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Contrast and verification of models\u003c/h2\u003e\u003cp\u003eTo validate the model, we used 2nd HDMU data as the training set and NSMU data as the test set to establish Logistic regression, classification and regression tree (CART), and Bayesian modeling. And compare the performance parameters between the models.\u003c/p\u003e\u003c/div\u003e"},{"header":"3 Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Basic information of patients\u003c/h2\u003e\u003cp\u003eAmong the 832 patients, 438 (52.64%) were treated at the 2nd HDMU, and 394 (47.36%) were treated at NHSMU; the youngest patient was 19 years old, the oldest was 96 years old, and the average age was 64 years old. In term\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003es\u003c/span\u003e of gender, 687 (82.6%) were males, while 145 (17.4%) were females. There were 482 (57.9%) solitary tumors and 350 (42.1%) multiple tumors. As for tumor pathology, there were 579 cases (69.6%) at Ta-T1 stage, 158 cases (19%) at T2 stage, 69 cases (8.3%) at T3 stage, and 26 cases (3.1%) at T4 stage. Among whole participants, there were 99 cases (11.9%) of low malignant potential, 284 cases of low grade (34.1%), and 449 cases of high grade (54%). In the study of treatment modalities, 539 (64.8%) cases received TURBT, 66 (7.9%) cases received partial cystectomy, and 227 (27.3%) cases received total cystectomy. The detailed comparison of the basic characteristics of the patient population in the two hospitals is shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Apart from gender, tumor number, and N stage, there were significant differences in other characteristics between the two hospitals, but the trend was consistent. The differences were considered to be influenced by regional economy, population culture, as well as regional radiation effects between the hospitals.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparison of basic characteristics of patients undergoing bladder cancer surgery in two hospitals\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eInfluencing factors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2nd HDMU(n\u0026thinsp;=\u0026thinsp;438)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNHSMU(n\u0026thinsp;=\u0026thinsp;394)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYears old\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e66.63\u0026thinsp;\u0026plusmn;\u0026thinsp;11.604\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e59.27\u0026thinsp;\u0026plusmn;\u0026thinsp;14.291\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003egender\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFemale (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e83(18.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e62(15.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMale (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e355(81.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e332(84.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.222\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTumor multiplicity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUnifocal (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e244(55.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e238(60.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMultifocal (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e194(44.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e156(39.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.17\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological Grade\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLMP (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4(0.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e95(24.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLow (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e158(36.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e126(32)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHigh (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e276(63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e173(43.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological T stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTa-T1(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e329(75.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e250(63.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eT2(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e65(14.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e93(23.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eT3(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e32(7.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e37(9.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eT4(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e12(2.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14(3.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.003\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological N stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN0(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e428(97.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e389(98.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN1(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6(1.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2(0.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN2(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4(0.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3(0.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.431\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSurgical method\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTURBT (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e304(69.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e235(59.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003epartial\u0026nbsp;cystectomy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e18(4.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e48(12.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eradical\u0026nbsp;cystectomy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e116(26.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e111(28.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3-year OS\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSurvival (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e385(87.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e364(92.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDeath (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e53(12.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e30(7.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.031\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eLMP: low malignant potential, TURBT: transurethral bladder tumor resect, T: Tumor, N: Lymph node staging., OS: Overall survival.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Extraction of data characteristics\u003c/h2\u003e\u003cp\u003eThe training group included 438 patients treated at the 2nd HDMU, and the data feature selection technique employed was univariate analysis. Patients were classified into survival and death groups based on 36-month survival after surgery. In univariate analysis, age, smoking, lymphocyte count, hemoglobin, albumin/globulin ratio, creatinine clearance rate, blood calcium, T stage, N stage, pathological grade, and surgical procedure differed significantly across groups (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05, Supplementary Table\u0026nbsp;1). These variables were fed into the DNN model. In addition, we included a history of hypertension or cardiovascular and cerebrovascular illnesses with a P-value of 0.079 (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), which, while not statistically significant, was included due to its potential clinical significance. At the same time, we conducted collinearity diagnosis on the included variables. The collinearity diagnosis results showed that the variance inflation factors of each variable were all less than 10, with a maximum of 2.861 and a minimum of 1.068, indicating no collinearity between the independent variables.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparison of general information between the death group and the survival group of BUC patients in the 2nd HDMU 3 years after surgery\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eInfluencing factors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003edeath group(n\u0026thinsp;=\u0026thinsp;53)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003esurvival group(n\u0026thinsp;=\u0026thinsp;385)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYears old\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e79.45\u0026thinsp;\u0026plusmn;\u0026thinsp;10.079\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e64.85\u0026thinsp;\u0026plusmn;\u0026thinsp;10.66\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003egender\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFemale (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e12(22.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e71(18.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMale (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e41(77.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e314(81.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.464\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension/ Cardio-cerebrovascular Disease\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNone (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e28(52.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e257(66.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e25(47.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e128(33.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.079\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ediabetes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNone (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e45(84.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e341(88.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8(15.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e44(11.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.357\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003esmoke\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNone (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e40(75.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e233(60.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e13(24.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e152(39.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.039\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWBC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eg/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6.47\u0026thinsp;\u0026plusmn;\u0026thinsp;1.188\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6.4\u0026thinsp;\u0026plusmn;\u0026thinsp;1.805\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.717\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNE\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eg/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4.11\u0026thinsp;\u0026plusmn;\u0026thinsp;1.107\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.91\u0026thinsp;\u0026plusmn;\u0026thinsp;1.514\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.341\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLYM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eg/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.74\u0026thinsp;\u0026plusmn;\u0026thinsp;0.539\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.95\u0026thinsp;\u0026plusmn;\u0026thinsp;0.673\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.035\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNE/LYM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.77\u0026thinsp;\u0026plusmn;\u0026thinsp;2.003\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.39\u0026thinsp;\u0026plusmn;\u0026thinsp;2.412\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.273\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eg/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e130.9\u0026thinsp;\u0026plusmn;\u0026thinsp;17.828\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e141.15\u0026thinsp;\u0026plusmn;\u0026thinsp;18.782\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eg/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e222.01\u0026thinsp;\u0026plusmn;\u0026thinsp;70.329\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e216.26\u0026thinsp;\u0026plusmn;\u0026thinsp;63.659\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.543\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eg/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e38.12\u0026thinsp;\u0026plusmn;\u0026thinsp;4.313\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e42.91\u0026thinsp;\u0026plusmn;\u0026thinsp;4.249\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB/GLO\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.44\u0026thinsp;\u0026plusmn;\u0026thinsp;0.357\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.66\u0026thinsp;\u0026plusmn;\u0026thinsp;0.367\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eU/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e23\u0026thinsp;\u0026plusmn;\u0026thinsp;13.368\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22.71\u0026thinsp;\u0026plusmn;\u0026thinsp;12.498\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.874\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGOT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eU/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e21.73\u0026thinsp;\u0026plusmn;\u0026thinsp;20.926\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23.36\u0026thinsp;\u0026plusmn;\u0026thinsp;16.709\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.520\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLDH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eU/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e198.66\u0026thinsp;\u0026plusmn;\u0026thinsp;39.742\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e198.66\u0026thinsp;\u0026plusmn;\u0026thinsp;37.189\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.253\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eU/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e75.67\u0026thinsp;\u0026plusmn;\u0026thinsp;26.263\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e71.92\u0026thinsp;\u0026plusmn;\u0026thinsp;29.553\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.381\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCCr\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eml/(min*1.73m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e71.99\u0026thinsp;\u0026plusmn;\u0026thinsp;19.398\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e81.98\u0026thinsp;\u0026plusmn;\u0026thinsp;14.196\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSerum Calcium\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003emmol/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.18\u0026thinsp;\u0026plusmn;\u0026thinsp;0.03\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.29\u0026thinsp;\u0026plusmn;\u0026thinsp;0.01\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePhosphorus\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003emmol/L\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.09\u0026thinsp;\u0026plusmn;\u0026thinsp;0.249\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.13\u0026thinsp;\u0026plusmn;\u0026thinsp;0.433\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.525\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTumor multiplicity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUnifocal (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e25(47.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e219(56.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMultifocal (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e28(52.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e166(43.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.246\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological Grade\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLMP (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4(1.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLow (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5(9.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e153(39.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHigh (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e48(90.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e228(59.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological T stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTa-T1(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e22(41.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e307(79.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eT2(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e16(30.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e49(12.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eT3(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e9(17.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23(6.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eT4(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6(11.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e6(1.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological N stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN0(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e47(88.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e381(99.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN1(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4(7.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2(0.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN2(%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2(3.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2(0.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSurgical method\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTURBT (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e27(50.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e277(71.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003epartial\u0026nbsp;cystectomy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10(18.9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8(2.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eradical\u0026nbsp;cystectomy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e16(30.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e100(26.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eWBC: White blood cell count, NE: neutrophilic granulocyte, LYM: lymphocyte, HGB: hemoglobin, PLT: platelet, ALB: albumin, GLO: globular proteins, ALT: glutamic-pyruvic transaminase, GOT: glutamic oxaloacetic transaminase, LDH: lactate dehydrogenase, ALP: alkaline phosphatase, CCr: Creatinine Clearance Rate, LMP: low malignant potential, TURBT: transurethral bladder tumor resect, T: Tumor staging, N: Lymph node staging\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.3 DNN model establishment\u003c/h2\u003e\u003cp\u003eIn the DNN modeling process, TensorFlow Keras framework was used, and the data result visualization used the TensorBoard modeling data visualization panel. Mapping the classification into three dimensions, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e shows the modeling effect, effectively distinguishing whether or not the patients died at the follow-up cut-off point. The convergence of each neural network layer during model training is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, indicating that the predicted value of the function closely fits the actual situation of the patients, with minima\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003el\u003c/span\u003e error and prediction bias, thereby demonstrating the strong predictive efficacy of our model.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe stochastic gradient descent algorithm was used to determine the relative importance of influencing factors. The results indicate that T stage has the highest influence on patient survival, followed by pathological grade, hypertension or cardiovascular and cerebrovascular diseases, hemoglobin concentration, serum calcium, smoking, serum albumin level, lymphocyte count, age, serum albumin/globulin ratio, surgical method, N stage, and creatinine clearance rate (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eAfter establishing the DNN model, we randomly select 88 cases of data from 20% of the training set as internal validation. And the accuracy of the model was validated using data from NHSMU patients with bladder cancer. The areas under the ROC curves of the models are 0.890 (95% CI 0.828\u0026ndash;0.951) for the training group (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e), 0.863 (95% CI 0.741\u0026ndash;0.985) for the internal validation group (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e), and 0.835 (95% CI 0.744\u0026ndash;0.926) for the external validation group (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e). The relevant ROC curves are shown in the figure. The fact that the accuracy rate of the model in the training group was 88.57% (95% CI: 0.8805\u0026ndash;0.8909), the accuracy rate of the model in the verification group was 92.05% (95% CI: 0.9168\u0026ndash;0.9242), and the accuracy rate for the entir\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ee\u003c/span\u003e sample population was 90.56% (95% CI: 0.9017\u0026ndash;0.9095). These results demonstrate that the DNN model has a high level of predictive performance in this instance. Additionally, a calibration curve of the DNN model was generated using Python (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e), which illustrates a good fit between the predicted patient survival and the actual condition.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eThe weight value of factors influencing postoperative survival and rank by importance\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003efactors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eimportance\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eT\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.5876238\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003epathological grading\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.5786078\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension/ Cardio-cerebrovascular Disease\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.5035198\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.4522362\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSerum Calcium\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.4269765\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003esmoke\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.3287318\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.3209476\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLYM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.2988171\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.2210308\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB/GLO\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.2157595\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSurgical method\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.1922349\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.1830217\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCCr\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.151636\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.4 Contrast and verification of models\u003c/h2\u003e\u003cp\u003eWe conducted traditional predictive model modeling using the same set of data to verify the predictive performance of the DNN model. Firstly, we applied the 2nd HDMU data for Logistic multivariate analysis based on the influencing factors selected through univariate analysis, and the results are shown in supplementary table 3. Then Logistic regression, classification and regression tree (CART), and Bayesian modeling were performed separately. And the data of NHSMU was applied for validation. We compared and analyzed the accuracy, precision, sensitivity, and specificity of these models. (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e) It can be found that the DNN model was significantly superior to the other three methods in terms of accuracy, sensitivity and F score. And the models showed similar specificity. The receiver-operating-characteristic (ROC) curve of the models was drawn in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e, and the area under curve (AUC) of the DNN model was 0.939, which was superior to the others.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eLogistic multivariate analysis based on the influencing factors selected through univariate analysis in the 2nd HDMU\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eInfluencing factors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u003cp\u003e95% C.I.for EXP(B)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLower\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eUpper\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.078\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.040\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.116\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;0.001\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHypertension/ Cardio-cerebrovascular Disease\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.550\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.811\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.963\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.185\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003esmoke\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.154\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.580\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.298\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.683\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLYM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.923\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.565\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.508\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.750\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.997\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.977\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.762\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.041\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.939\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.154\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.444\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALB/GLO\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.805\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.354\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.831\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.605\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCCr\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.996\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.977\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.016\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.696\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSerum Calcium\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.241\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.015\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.910\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.317\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological Grade\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.214\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.610\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.419\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.581\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological T stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1.618\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.979\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.674\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.061\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePathological N stage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e2.691\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.258\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5.757\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.011\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSurgical method\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.656\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.408\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.055\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.082\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparison of the Logistic regression model, the classification and regression tree model, the Bayesian model, and the DNN model\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSensitivity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eF Score\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.91\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.61\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.90\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.94\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.73\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLogistic regression\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.82\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.63\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.17\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.97\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.26\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClassification and regression tree\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.84\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.92\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.14\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.99\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.25\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBayesian\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.81\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.51\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.95\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.29\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"4 Discussion","content":"\u003cp\u003eBUC is a common cancer with a high recurrence rate, and tumor pathological stage and grade are the most independent prognostic predictors for survival\u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. Recent years, through basic experiments and bioinformation analysis, some glycoproteins in urine, gene mutation sites in genetic testing, and prognostic markers based on transcriptome files have been found to be associate\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ed\u003c/span\u003e with the proliferation, invasion, and metastasis of tumor cells, and can predict the clinical efficacy of treatment and the survival of patients. These indicators have accurate predictions, stable efficacy, and strong individual specificity. However, due to the limitations in detection technology and cost, they have not been popularized in clinical practice. Additionally, the prognostic value of clinical routine indicators for patients' conditions has been explored. For instance, serum albumin, serum albumin/globulin ratio, granulocyte/lymphocyte ratio have been identified as indicators that can be widely used in clinic without imposing additional economic burdens on patients. However, most of the current reports on survival prognosis are single-center, small-sample studies that used Logistic or Cox univariate/multivariate regression models for analysis. The conclusions obtained were inconsistent. Logistic and Cox models are generalized linear models. While few studies used non-linear artificial intelligence analysis to conduct interaction analysis of influencing factors and weight quantification. Theoretically, non-linear models have better predictive performance than linear models. Therefore, applying non-linear artificial intelligence analysis can comprehensively evaluate various prognostic indicators and establish a prognostic model with high accuracy and good universality. Such a model can be used to quantitatively evaluate the survival prognosis of patients with different tumor burdens after receiving various surgical treatments, which would be highly desirable.\u003c/p\u003e\u003cp\u003eNeural network technology, originated in the 1950s, has an input layer, an output layer, and a hidden layer. Through the calculation of the hidden layer, each variable receives a weight in the hidden layer, and the final change result is sent to the output layer to generate the prediction result. Nowadays, several researchers are attempting to apply artificial intelligence technologies for disease diagnosis and therapy modeling. In the field of urological tumors, artificial intelligence technology has made many advances in improving the diagnosis and treatment of prostate cancer, kidney cancer, and bladder cancer. Treatment plan optimization and patient follow-up education are not well-researched and the technology is still in its infancy\u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e. This study uses the DNN algorithm to model, mainly relying on the method of backpropagation to improve model fitting. Through repeated training, the results obtained each time are compared with the expected results for error analysis. Subsequently, the weight and threshold of each neural node are adjusted based on the comparison results, enabling the model to progressively approach the expected result and improve its accuracy. In the end, the test accuracy rate of the model training group in this example was 88.57%, and the test accuracy rate of the external verification group was 92.05%, and the calibration curve shown that the patient survival predicted by the DNN model has a good fit to the actual condition, which proves that the model has good predictive performance, and subsequent software development can be performed according to the weight. We also used the same data for Logistic regression, CART, and Bayesian modeling separately. Through comparing and analyzing the DNN model with the other models, we\u0026rsquo;ve found that the DNN model is significantly superior to other three methods in F score, and ha\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003es a\u003c/span\u003e better AUC than others. This indicates that the DNN model in this study has better prediction performance for the long-term survival of BUC patients after surgery and is suitable for clinical reference.\u003c/p\u003e\u003cp\u003eWith the exception of gender, tumor number, and N stage, there were significant differences in the other characteristics of the bladder cancer patients in the two hospitals in this study. The reasons were considered to be the economic and cultural level of the population between regions, as well as the influence of hospitals. That is to say, people in the developed area (Guangzhou) pay more attention to their own physical condition, undergo regular physical examination and seek medical treatment at the early stage of the disease (such as the first occurrence of hematuria). As a result, the average age of the patient population is younger, and the pathological grade is lower. At the same time, patients' higher economic level and higher requirements for quality of life promote the preference for more influential provincial hospitals. Therefore, the T stage of bladder cancer patients in NHSMU is higher, and the proportion of partial cystectomy and total cystectomy is higher than that in the 2nd HDMU. Although there are statistical differences in the basic characteristics of patients between regions, the trend of the characteristics is consistent and the difference is not large. The data of patients from the north and the south were used for modeling and verification, which can verify the generality of the model obtained for Chinese bladder cancer patients.\u003c/p\u003e\u003cp\u003eIn this work, the 3-year OS was utilized as the research endpoint for univariate analysis, and the indicators with statistically significant differences between survival and death patients were incorporated into the DNN model for analysis. Age is the fundamental metric that indicates physical reserve, and increased age is a risk factor for the short- and long-term survival of nearly all malignant tumor diseases\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. The history of hypertension or cardiovascular and cerebrovascular illnesses reveals the patient's underlying illness, which can considerably impact the patient's risk of mortality from all causes\u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. Smoking also is a clear risk factor for bladder cancer recurrence\u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e. The current study provides evidence that vitamin D deficiency may be related to tumor recurrence and progression, and blood calcium level can indirectly reflect the vitamin D level in patients; therefore, low calcium may be a risk factor affecting the survival of cancer patients\u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e. Tumor pathological T stage, N stage, and grading reflect the tumor burden and malignancy, the higher the stage and grading, the worse the prognosis\u003csup\u003e[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e. The level of hemoglobin, serum albumin, and albumin/globulin ratio can reflect the nutritional level of the patient and the degree of tumor consumption. Malnutrition is considered to be the main reason for the high incidence of postoperative complications, which can lead to the weakening of the patient's defense mechanism. Multiple studies have confirmed that anemia and hypoalbuminemia are adverse prognostic factors for patients requiring surgery. Complications, and rapid cancer progression are more common in these patients\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e. Lymphocyte count reflects the immune ability of the body, and studies have found that the decline of the body's immune ability is related to the poor prognosis of tumors\u003csup\u003e[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e. The creatinine clearance rate reflects the patient's renal function and the level of basic physical quality, making it one of the predictors of survival prognosis. The above factors were confirmed to be related to the survival and prognosis of BUC patients through univariate analysis and were incorporated into the DNN model for further analysis.\u003c/p\u003e\u003cp\u003eIn this case, the DNN model analysis found that the top three factors affecting the patient's surgical prognosis are: tumor T stage, pathological grade, and hypertension or cardiovascular and cerebrovascular diseases, which have a higher predictive effect on the long-term survival of patients than other indicators. The higher the pathological stage and tumor grade, the worse the underlying disease state, and the worse the prognosis of patients after surgery, which is consistent with the previous research results. Other influencing factors, in descending order of importance, were hemoglobin concentration, serum calcium, smoking, serum albumin level, lymphocyte count, age, serum albumin/globulin ratio, surgical method, N stage, and creatinine clearance rate. Different statistical methods may lead to differences in the calculation results of weights. It can be seen that the Logistic and DNN model diffe\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003er\u003c/span\u003e somewhat in the weight ranking of the influencing factors. The influencing factor with the highest OR value obtained from Logistic analysis is N stage, followed by T stage, hypertension or cardiovascular and cerebrovascular diseases, and pathological grade, which is different from the DNN model. In clinical practice, T stage, N stage, and pathological grade are all important influencing factors that significantly affect the choice of surgical method and postoperative survival of patients. Regardless of the T stage and pathological grade of the patient, the presence of lymph node metastasis indicates poor prognosis. The survival time of patients in N1 or N2 stage is significantly shorter than N0 stage. However, only a few patients with advanced BUC have lymph node metastasis. In this study, only 27(3.2%) patients were classified as N1 or N2 in N stage, and the 3-year mortality rate was as high as 29.6% (8 cases). In clinical practice, clinicians pay more attention to the T stage and pathological grading of N0 BUC patients. Therefore, T stage and pathological grade may have greater prognostic value than N stage for the overall BUC patients.\u003c/p\u003e\u003cp\u003eTherefore, we believe that the weight ranking of influencing factors using the DNN model is closer to clinical practice compared to the Logistic model.\u003c/p\u003e\u003cp\u003eThe number of follow-up cases in this study is large, and it is a multi-center study. The follow-up data of NHSMU were included for external verification. Through external verification, the DNN model which has an accuracy rate of 92% in predicting the 3-year OS of BUC patients, proves that the model in this study has good predictive performance. According to the corresponding weight of each prognostic factor, the software is being developed. By inputting the value of each influencing factor, the corresponding 3-year all-cause survival rate of the patient can be obtained, and the difference in the long-term survival rate after different surgical procedures can be calculated. This model has the characteristics of high accuracy and strong quantitative comparison, and it shows excellent application prospects in clinical work.\u003c/p\u003e"},{"header":"5 Limitations and future prospectives","content":"\u003cp\u003e1.This article is a preliminary exploration of the DNN artificial intelligence algorithm in the survival and prognosis of BUC surgery. The source of data is limited, and there may be differences in the baseline conditions of patients and the technical level of doctors. 2. All patients did not receive adjuvant or neoadjuvant therapy, but the adjuvant therapy plan can be included as one of the influencing factors for artificial intelligence learning to improve the survival prognosis model in further studies. In the future, the research team will use the self-learning nature of the DNN artificial intelligence model to incorporate patient data from more centers. They will enrich the prognostic factors, adjust the weight of the model neuron unit, improve the prediction accuracy of the model, and increase its survival prediction efficiency, thereb\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ey\u003c/span\u003e providin\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eg\u003c/span\u003e different surgical options for patients. The intuitive preoperative prediction of the treatment effect facilitates preoperative communication between doctors and patients, enabling them to select appropriate treatment methods and formulate individualized follow-up plans.\u003c/p\u003e"},{"header":"6 Conclusions","content":"\u003cp\u003eUsing DNN modeling and external validation, BUC patients' postoperative survival variables can be predicted. This can predict bladder cancer surgery outcomes before surgery. Additionally, this model can help patients choose surgery and postoperative follow-up techniques using algorithms.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eYang Yu is the guarantor. Yue Zhang, Ce Zhang, and Lijie Wen conceived, designed, and collected the questionnaire for the patient. Bo Yang and Xiang Gao collated data and produced graphs. Yue Zhang and Ying Ke drafted the article and embellished it with revisions. Yang Yu provided funding support. All authors made substantial contributions to the article draft and critically revised it. All authors approved the submitted and final versions.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eThis study was supported by the \u0026ldquo;1+X\u0026rdquo; program for Cross-disciplinary Innovation Projects of the Second Hospital of Dalian Medical University (2022JCXKZD05), In-Hospital Training Fund of the Second Hospital of Dalian Medical University (dy2yynpy202220), Liaoning Provincial Department of Education 2021 Scientific Research Funding Project (LJKZ0873), the Second Hospital of Dalian Medical University - Dalian Institute of Chemical Physics, Chinese Academy of Sciences \u0026quot;Collaborative Innovation Centre for Individualized Diagnosis and Treatment\u0026quot; Jointly Funded Project (UF-ZD-202014).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAcknowledgments\u003c/p\u003e\n\u003cp\u003eWe thank all the participants who were willing to be followed up during this period.\u003c/p\u003e\n\u003cp\u003eEthics approval and consent to participate\u003c/p\u003e\n\u003cp\u003eAll procedures performed in this study were in accordance with the ethical standards of the Declaration of Helsinki. Approval was granted by the ethics committee of the Second Hospital of Dalian Medical University (Approval No. 177, 2023). Informed consent was obtained from all individual participants included in the study.\u003c/p\u003e\n\u003cp\u003ePatient and Public Involvement\u003c/p\u003e\n\u003cp\u003ePatients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.\u003c/p\u003e\n\u003cp\u003eConflict of Interest\u003c/p\u003e\n\u003cp\u003eAll authors have no competing interest to declare.\u003c/p\u003e\n\u003cp\u003edata availability statement\u003c/p\u003e\n\u003cp\u003eAll data generated or analyzed during this study are included in this published article.\u003cstrong\u003e\u003cbr\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSIEGEL RL, MILLER K D, FUCHS H E et al. Cancer statistics, 2022[J/OL]. CA: A Cancer Journal for Clinicians, 2022, 72(1): 7\u0026ndash;33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3322/caac.21708\u003c/span\u003e\u003cspan address=\"10.3322/caac.21708\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXIA C, DONG X, LI H, et al. Cancer statistics in China and United States, 2022: profiles, trends, and determinants[J/OL]. Chin Med J. 2022;135(5):584\u0026ndash;90. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/CM9.0000000000002108\u003c/span\u003e\u003cspan address=\"10.1097/CM9.0000000000002108\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGONTERO P. European Association of Urology Guidelines on Non\u0026ndash;muscle-invasive Bladder Cancer (TaT1 and Carcinoma In Situ)\u0026mdash;A Summary of the 2024 Guidelines Update[J]. invasive Bladder Cancer.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eANTAR R M, XU V E, FARAG C M, et al. Oncologic outcomes of neoadjuvant chemotherapy and lymph node dissection with partial cystectomy for muscle-invasive bladder cancer[J/OL]. Translational Androl Urol. 2024;13(8). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21037/tau-24-165\u003c/span\u003e\u003cspan address=\"10.21037/tau-24-165\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHU J, CHEN J, OU Z, et al. Neoadjuvant immunotherapy, chemotherapy, and combination therapy in muscle-invasive bladder cancer: A multi-center real-world retrospective study[J/OL]. Cell Rep Med. 2022;3(11):100785. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.xcrm.2022.100785\u003c/span\u003e\u003cspan address=\"10.1016/j.xcrm.2022.100785\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYUE Z, FAN FEIL. Prognostic value of preoperative serum albumin in patients with non-muscle-invasive bladder cancer undergoing transurethral resection of bladder tumor [J/OL]. J South Med Univ. 2018;38(2):192. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3969/j.issn.1673-4254.2018.02.12\u003c/span\u003e\u003cspan address=\"10.3969/j.issn.1673-4254.2018.02.12\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEpidemiology LOBON. Screening, and Prevention of Bladder Cancer[J/OL]. 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.euo.2022.10.003\u003c/span\u003e\u003cspan address=\"10.1016/j.euo.2022.10.003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOFNER H, LAUKHTINA E HASSLERMR, et al. Blood-Based Biomarkers as Prognostic Factors of Recurrent Disease after Radical Cystectomy: A Systematic Review and Meta-Analysis[J/OL]. Int J Mol Sci. 2023;24(6):5846. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ijms24065846\u003c/span\u003e\u003cspan address=\"10.3390/ijms24065846\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFERNANDEZ-PRADO R. Creatinine Clearance Is Not Equal to Glomerular Filtration Rate and Cockcroft-Gault Equation Is Not Equal to CKD-EPI Collaboration Equation[J/OL]. Am J Med. 2016;129(12). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.amjmed.2016.08.019\u003c/span\u003e\u003cspan address=\"10.1016/j.amjmed.2016.08.019\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWENHAO X, XI T, Aihetaimujiang A, et al. A systematic review of current advancements of artificial intelligence in genitourinary cancers [. J] China Oncol. 2022;32(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.19401/j.cnki.1007-3639.2022.01.009\u003c/span\u003e\u003cspan address=\"10.19401/j.cnki.1007-3639.2022.01.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCHANGFA X. Fractions and trends of cancer burden attributable to population ageing in China [J]. Chin J Oncol. 2022;44(1):79\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBRAY F, LAVERSANNE M. The ever-increasing importance of cancer as a leading cause of premature death worldwide[J/OL]. Cancer. 2021;127(16):3029\u0026ndash;30. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/cncr.33587\u003c/span\u003e\u003cspan address=\"10.1002/cncr.33587\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLISHA L, HANG HANGHANGL. An analysis of disease burden of prostate, bladder and kidney cancers attributable to smoking in China from 1990 to 2019 [J]. Chin J Evidence-Based Med. 2022;22(05):530\u0026ndash;6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7507/1672-2531.202201059\u003c/span\u003e\u003cspan address=\"10.7507/1672-2531.202201059\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKEUM N, LEE D H, GREENWOOD D C, et al. Vitamin D supplementation and total cancer incidence and mortality: a meta-analysis of randomized controlled trials[J/OL]. Ann Oncol. 2019;30(5):733\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/annonc/mdz059\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdz059\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBoot IWA, Wesselius A, Yu EYW, et al. Dietary vitamin D intake and the bladder cancer risk: A pooled analysis of prospective cohort studies. Clin Nutr. 2023;42(8):1462\u0026ndash;74. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.clnu.2023.05.010\u003c/span\u003e\u003cspan address=\"10.1016/j.clnu.2023.05.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVAN BRUWAENE S, COSTELLO A J, VAN POPPEL H. Prognosis of node-positive bladder cancer in 2016[J]. Minerva Urologica E Nefrologica\u0026thinsp;=\u0026thinsp;The. Italian J Urol Nephrol. 2016;68(2):125\u0026ndash;37.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGRIMM T, BUCHNER A, SCHNEEVOIGT B, et al. Impact of preoperative hemoglobin and CRP levels on cancer-specific survival in patients undergoing radical cystectomy for transitional cell carcinoma of the bladder: results of a single-center study[J/OL]. World J Urol. 2016;34(5):703\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00345-015-1680-7\u003c/span\u003e\u003cspan address=\"10.1007/s00345-015-1680-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang S, Du J, Zhong X, et al. The prognostic value of the systemic immune-inflammation index for patients with bladder cancer after radical cystectomy. Front Immunol. 2022;13:1072433. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fimmu.2022.1072433\u003c/span\u003e\u003cspan address=\"10.3389/fimmu.2022.1072433\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Published 2022 Nov 29.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Bladder cancer, Deep learning algorithm, Surgery prognosis system","lastPublishedDoi":"10.21203/rs.3.rs-7937665/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7937665/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjective\u003c/h2\u003e\u003cp\u003eTo discover the variables that affect bladder cancer (BC) patients' survival and prognosis after surgical treatment, and to use this knowledge to build an artificial intelligence (AI)-based recommendation algorithm.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eThis study comprised 832 BC patients who underwent surgery at The Second Affiliated Hospital of Dalian Medical University (2nd HDMU) and Nanfang Hospital of Southern Medical University (NHSMU) between January 2007 and January 2019. Their clinical and follow-up data were obtained. The 2nd HDMU patients were the training group, whereas NHSMU patients were the test group for external validation. An AI algorithm model was created using the deep neural network (DNN). The parameters influencing patient survival were analyzed and ranked with the assistance of AI algorithm.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eOut of the 832 bladder cancer patients included in this study, 438 (52.64%) were treated in the 2nd HDMU, while 394 (47.36%) were in the NHSMU. Among the BC cases, 579 (69.6%) were diagnostic of non-muscle invasive bladder cancer, while only 253 (30%) were muscle-invasive bladder cancer. In terms of surgical intervention, 539 (64.8%) patients underwent transurethral resection of bladder tumor, 66 (7.9%) received partial cystectomy, and 227 (27.3%) received total cystectomy. We concluded that the factors affecting the survival and prognosis of patients, in descending order, were T stage, pathological grade, hypertension or cardiovascular and cerebrovascular diseases, hemoglobin concentration, serum calcium, smoking, serum albumin level, lymphocyte count, age, serum albumin/globulin ratio, surgical method, N stage, and creatinine clearance rate. The testing group evaluated and confirmed this model to predict BC patients' survival before surgery.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eUtilizing DNN modeling and external validation, the influencing factors of postoperative survival can be predicted for patients with BC. It can be employed to forecast BC patients' surgical outcomes before surgery. Additionally, this model can provide algorithmic assistance in selecting surgical and postoperative follow-up strategies for such patients.\u003c/p\u003e","manuscriptTitle":"Prediction of survival and prognostic factors in patients with bladder cancer after surgery using artificial intelligence recommendation algorithm: a preliminary study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-01 08:10:26","doi":"10.21203/rs.3.rs-7937665/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"1e7d4e86-3209-4ada-9fca-b45240d55c91","owner":[],"postedDate":"December 1st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-03T15:25:43+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-01 08:10:26","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7937665","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7937665","identity":"rs-7937665","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}