Evaluation of Inflammatory Markers in Survival Analysis of Patients Undergoing Radical Cystectomy Using Machine Learning

Evaluation of Inflammatory Markers in Survival Analysis of Patients Undergoing Radical Cystectomy Using Machine Learning | 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 Evaluation of Inflammatory Markers in Survival Analysis of Patients Undergoing Radical Cystectomy Using Machine Learning NACİ BURAK ÇINAR, HASAN YILMAZ, EFE YILMAZ TAŞYÜREK, MELTEM KURT PEHLIVANOĞLU, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6717746/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 21 Oct, 2025 Read the published version in World Journal of Urology → Version 1 posted 13 You are reading this latest preprint version Abstract Background We aimed to create a Machine learning (ML) model using patient demographic, clinical and pathological data for prediction of overall survival in patients treated with radical cystectomy (RC). Secondly, we evaluated whether inflammatory markers contributed to this model. Methods We conducted a retrospective analysis of the institutional cystectomy database and identified consecutive RC patients. Dataset-1 (DS-1) was analyzed in ML models using 30 original features (including the target feature) encompassing preoperative, intraoperative, and postoperative data of the patients. All derived inflammatory markers were cumulatively added to DS-1 to create DS-2, and to test the specific contribution of inflammatory markers, they were systematically integrated in an ordinary order based on their predictive ability (DS-3). Markers without predictive contribution were excluded from the DS-3 model. In addition, the Shapley Additive Explanations (SHAP) method was used to examine the importance of each clinical feature and inflammatory marker. Results The median age of the 241 patients was 65 years. The mortality rate was 60.2% (145/241). Two and 5-year overall survival (OS) rates were 54.7% and 37.2%, respectively. According to DS-1, F1 scores were between 0.72–0.78. RandomForests and XGBoost models achieved the highest score of 0.78. DS-2 including all inflammatory markers but no significant improvement was found in F1 scores (0.73–0.78). In DS-3, firstly adding the systemic inflammatory response index (SIRI) to the original features and then neutrophil/lymphocyte ratio (NLR) and platelet/lymphocyte ratio (PLR), achieved the highest F1 score (0.80) in the Random Forest model. SHAP analyses showed that Tumor (T) stage, preoperative albumin and presence of lympho-vascular invasion (LVI) contributed most to model predictivity. Conclusions ML models derived using demographic/clinical features resulted in a maximum F1 score of 0.78. However, adding the most predictive inflammatory markers in the sequence SIRI, NLR and PLR to demographic data achieved the highest F1 score of 0.80. Furthermore, T stage and preoperative albumin were the strongest predictive factors in the ML models. Machine Learning Inflammatory markers Radical Cystectomy Shapley Additive Explanations (SHAP) Artificial intelligence Figures Figure 1 Introduction Bladder cancer is the 11th most common cancer worldwide, with approximately 550,000 new cases diagnosed annually(1), and it is a significant cause of morbidity and mortality(2). Muscle-invasive bladder cancer (MIBC) accounts for 25% of cases, with radical cystectomy (RC) and pelvic lymph node dissection as the standard treatment. However, despite the radical treatment modalities, the 5-year overall survival rate remains at 60%(3). Thus, to determine the optimal treatment strategy for improvement of patient survival rates, predictive models may be beneficial. There are existing models for prediction of post-cystectomy survival but an ideal model for this purpose has not been established in the literature, yet. Machine learning (ML) has emerged as a powerful tool in medicine, particularly in oncology, where it facilitates survival prediction and risk stratification. ML models have demonstrated significant advances in predictive models for outcomes. In the context of survival after RC, ML algorithms can simply analyse a wide range of patient data, including demographic data, clinical and laboratory results, and imaging findings. In addition, ML algorithms may also identify predictive effects of novel markers on patient survival. Thus, ML algorithms can further contribute to specialists in pre-operative risk management and the future course of the disease. Moreover, ML algorithms have the potential to become sustainable by adapting to new patient data. Inflammation plays an important role in tumor development and progression (4). Systemic inflammatory response cytokines are induced by the tumor microenvironment and this has led to the use of inflammatory markers as clinical and pathological predictive and prognostic factors in many studies(5). Numerous markers were found to be predictive for bladder cancer in the literature. However, a ML-based study concluded that the systemic inflammatory response (SIR) biomarker panel did not significantly improve the predictive ability of oncological outcomes in patients undergoing RC (6). In the present study, the aim was to create novel ML models using patient demographic, clinical and pathological data in patients undergoing RC. A secondary aim was to evaluate whether inflammatory markers contributed to these models. Materials and Methods This study included consecutive patients who underwent open RC, bilateral pelvic lymph node dissection, and urinary diversion (ileal loop or Studer pouch) for bladder cancer between 2009 and 2021. Demographic and clinic data of patients including preoperative, intraoperative, and postoperative information were analyzed using 30 original features (target feature included) in ML models (DS-1). Oncological outcomes for this study included time to death (overall survival). Preoperative blood parameters were obtained one day before the RC procedure. Numerous inflammatory markers were derived from these parameters. These markers included the systemic inflammatory response index (SIRI) (7), systemic immuno-inflammation index (SII)(8), neutrophil-to-lymphocyte ratio (NLR) (9), monocyte-to-lymphocyte ratio (MLR) (10), and platelet-to-lymphocyte ratio (PLR)(11). To more clearly evaluate the contribution of inflammatory markers, the DS-1 dataset was systematically expanded by adding inflammatory markers to create the DS-2 dataset. The predictive effectiveness of five derived inflammatory markers was evaluated based on the outcome of the death feature. To better assess the impact of inflammatory markers, we constructed several data subsets to explain the mortality feature using five specific inflammatory markers. We then recorded feature importance rankings derived from leading machine learning models, including CatBoost, XGBoost, and Random Forest. Based on these rankings, we systematically expanded the original dataset by sequentially incorporating markers according to their relative importance. The marker with the highest importance was initially added to the baseline feature set. Subsequently, the remaining markers were introduced individually, in descending order of importance. If the inclusion of a marker did not enhance the F1 score, it was excluded from the cumulative addition process. Ultimately, the highest performance scores were achieved using the Random Forest model. This approach aimed to evaluate the stepwise effect of each marker and determine the optimal combination of features for survival prediction, resulting in the DS-3 dataset. Supplementary Fig. 1. illustrates the feature importance ranking generated by the Random Forest model for the subset designed to explain the mortality feature using the five inflammatory markers The performance of applied ML algorithms was evaluated using F1-Score, Precision, Recall, Accuracy, and area under the curve-receiver operator characteristics (AUC-ROC) curve metrics(12). Machine Learning Algorithms This section summarizes the classification models and the Shapley Additive Explanations (SHAP) method used for feature selection. Supervised ML models, including Support Vector Machines (SVM), Multi-Layer Perceptron (MLP), Logistic Regression, Decision Trees, and ensemble learning models such as Random Forest were used(13). In addition, Gradient Boosting Machines (GBM), eXtreme Gradient Boosting (XGBoost)(14), Light Gradient Boosting Machine (LightGBM)(15), and Category Boosting (CatBoost) (16) were also applied. The rationale for the selection of these elements was as follows. Logistic Regression models linear relationships, while Decision Trees simplifies complex datasets into decision rules. Random Forest combines multiple decision trees, and Gradient Boosting algorithms (XGBoost, LightGBM, and CatBoost) enhance classification performance by combining weak learners. XGBoost excels in scalability, CatBoost optimizes categorical features with faster convergence, and LightGBM processes large datasets efficiently by growing trees vertically. SVM separates data using hyperplanes, and MLP is a neural network foundational to deep learning. The sensitivity of classification models to dependent variables can be investigated using SHAP values (17). The SHAP method is capable of assigning a degree of importance for each of the input parameters separately. In the present study, SHAP values were determined to examine the importance of the current clinical features and the importance of additional inflammatory markers. Evaluation Metrics Model performance was evaluated across five metrics: Accuracy, Precision, Recall, F1-score, and ROC-AUC. Considering the class imbalance, the primary performance metric was selected as the F1-score, which balances Precision and Recall metrics. Grid Search-CV and K-Fold Cross-Validation methods were employed to determine the optimal hyperparameters of the models and enhance their overall performance. During the hyperparameter optimization process, various parameter combinations were systematically evaluated, and each combination was tested using 10-fold cross-validation. This approach enabled the identification of the parameters that yielded the highest performance for each model. Once the optimal parameters were determined, a final 10-fold cross-validation was conducted on the entire dataset to obtain the ultimate performance metrics. For this, the dataset was randomly divided into ten equal parts, and each model was trained on nine of these subsets while being tested on the remaining one. This process was repeated ten times, ensuring each subset was a test set once. Applying this technique, the model underwent a comprehensive evaluation across the entire dataset. This primary objective of this approach was to ensure that the models performed well on a specific patient group and to guarantee a generalizable performance, making the models robust across different data distributions. Statistical analyses The Kolmogorov-Smirnov test was used to determine the normality of the numeric data. Categorical variables are expressed as count/frequency (n) and percentage (%), and continuous variables are presented as mean ± standard deviation (SD) or, if not normally distributed, median and interquartile range (IQR). The Kaplan-Meier method was used for OS calculations, using SPSS, version 21 (IBM Inc., Armonk, NY, USA) Results A total of 251 patients undergoing RC were identified over the study period. Patients who had previously received radiotherapy (n = 10) were excluded, resulting in a total of 241 patients. The median age was 65 years. Approximately half of the patients presented with advanced clinical and pathological T stage disease (T3 or T4). The median (IQR) follow-up duration was 20 (9–52) months. The mortality rate was 60.2% (145/241). Two and 5-year overall survival (OS) rates were 54.7% and 37.2%, respectively. The median OS was 30.00 months (95% CI 15.06–44.93). The demographic and clinical characteristics of the patients are presented in Table 1. The performance of nine different ML models was thoroughly evaluated over DS-1 (Table 2 ). These analyses resulted in an F1 score between 0.72–0.78. Random Forests and XGBoost models achieved the highest score of 0.78. All inflammatory markers were cumulatively added to the DS-1 to create DS-2, however no significant improvement was observed in F1 scores, ranging from 0.73 to 0.78 on subsequent performance analysis. To investigate the effect of inflammatory markers more deeply, the DS-1 dataset was systematically expanded by adding inflammatory markers in an order based on the feature importance scores obtained from the prominent models CatBoost, XGBoost, and RandomForests algorithms to create DS-3. Initially, the most predictive inflammatory marker was added to the original feature set alone. Then, other derived markers were added in decreasing order according to their predictive value. While following this sequence, if a specific inflammatory marker did not increase the F1-score metric, it was removed from this cumulatively advanced addition. The sequence was: Original + SIRI, Original + SIRI + LMR, Original + SIRI + NLR, Original + SIRI + NLR + PLR, and Original + SIRI + NLR + PLR + SII. In these subvariants, sequential addition of SIRI, NLR and PLR to the original features in the Random Forest achieved the highest F1 score of 0.80 (Table 3). Figure 1 -A, Fig. 1 -B, and Fig. 1 -C present the explainability analysis of the Random Forest model using SHAP values on different datasets. These figures are designed to assess the impact of various features on the model's performance in predicting survival. In the SHAP analyses of DS-1, T stage, preoperative albumin, and lympho-vascular invasion (LVI) were identified as the most influential factors in model classification performance (Fig. 1 A). In the DS-2, again T stage, preoperative albumin, and LVI were identified as the most influential factors, while SIRI and LMR were observed as the most effective markers among derived markers (Fig. 1 B). In DS-3, T stage, preoperative albumin, and preoperative platelet count were observed as the most effective markers seen as the most important factors, and the next most important was LVI (Fig. 1 C). Moreover, the SIRI, PLR, and NLR variables had positive effects on the performance of the model, with NLR and SIRI ranked especially highly. Discussion Currently, ML has increasingly contributed to the diagnosis and treatment of various diseases (18). Bladder tumors have increasingly been included in these studies In the literature review for the present research, there were a few studies related to bladder cancer and ML. Schuettfort et al analysed predictive factors in patients who underwent RC (6). Notably, their study which was the first ML-based study assessing the prognostic value of inflammatory markers focused specifically on the systemic inflammatory response (SIR) biomarker panel. This panel included the albumin-globulin ratio, NLR, De Ritis ratio, MLR, and the modified Glasgow prognostic score. The authors concluded that the created models including SIR biomarkers did not significantly improve the predictive ability of oncological outcomes(6). In the current study we found important differences to the study of Schuettfort et al. First, we comprehensively evaluated the performance of nine different ML models. Second, we added the inflammatory markers to the ML models in a specific sequence starting with the one with the highest contribution. Like the previous study, including all derived markers in a cumulative fashion did not improve the F1 scores. However, integrating the inflammatory markers with the most effective first to original data set, that is sequential addition of SIRI, NLR and PLR, achieved the highest F1 score of al combinations. This ML model was the most reliable model for survival prediction in our cohort. Of note, the SII and LMR markers made no contribution to model improvementThis is the first published study to demonstrate a significant improvement in performance of ML models with inflammatory marker data in the prediction of bladder cancer survival after RC. Furthermore, our findings align with our previous study based on the same dataset(19). In that study, multivariate Cox regression analyses demonstrated that both categorical and continuous SIRI were independent predictors of recurrence-free survival and overall survival. In another study, ML strategies were employed to assess the influence of lifestyle and occupational risk factors on bladder cancer(20). Wu et al. identified significant results using different deep convolutional neural network (DCNN) models to predict the response of bladder lesions to chemotherapy(21). Furthermore, an ML-based decision support model was developed in a large cohort to enhance the reliable and effective management of patients at high risk for postoperative complications following cystectomy and to optimize discharge planning (22). Tsai et al. analyzed five models incorporating calcium, alkaline phosphatase, albumin, urine ketones, occult hematuria, and liver transaminase data to differentiate bladder cancer from renal cancer, cystitis, and prostate cancer. Furthermore, ML has demonstrated potential in the non-invasive staging of bladder cancer through 3D texture analysis of bladder wall CT/MRI images, particularly in cases with challenging diagnostic features in differentiating malignant areas from normal urothelium during cystoscopy (23). Brieu et al. used a Random Forest model to accurately quantify tumor budding in muscle-invasive bladder cancer patients through immunofluorescence imaging, presenting it as a prognostic tool which appeared superior to TNM staging (24). In contrast, Galsky et al evaluated the effectiveness of treatment strategies for bladder cancer with clinical evidence of regional lymph node involvement (25). Then, ML models were developed to preoperatively predict lymph node metastasis, a known poor prognostic factor for survival with a 14-variable SVM model achieving the best performance(26). In the present study, SHAP analysis identified T stage, preoperative albumin, preoperative platelet count and the presence of LVI as having the most contribution to model prediction power in the Random Forest model. The SHAP analysis of three different models revealed similar key features. The results we obtained are consistent with previous studies in the literature (27–29). The prominence of NLR and SIRI in the most reliable model we developed shows that the derived markers not only make a significant contribution to our model but should also be considered in other modelling exercises for prognostic markers in bladder and other cancers. Limitation This study has some important limitations. First, the inclusion of a large number of features derived from patient data, coupled with a limited sample size, presents challenges in developing a robust ML model. Nevertheless, we anticipate that as the patient cohort expands, the predictive power of our model will improve, thereby yielding more reliable results. Furthermore, the retrospective design of the study constitutes another limitation. We were not able to use cancer-specific survival as an outcome variable because of insufficient data. Using overall survival instead may have affected our results. Conclusion We evaluated survival prediction in patients with bladder tumors using nine different ML models. We found that T stage, preoperative albumin and LVI contributed the most to accuracy of these models. In addition, adding the inflammatory markers SIRI, NLR and PLR to demographic data resulted in the highest F1 score. These findings demonstrate that inflammatory markers can contribute positively to ML models in this disease. However, more studies with larger patient numbers are needed to increase the predictive power of such ML models. Declarations Ethical Approval: This retrospective study was approved by the local ethics committee of Kocaeli University (Approval number: KÜ GOKAEK-2025/148). The study was conducted in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Consent to Participate: Written informed consent was obtained from all patients prior to surgery. Consent for Publication: Not applicable. Competing interests The authors did not receive support from any organization for the submitted work. The authors have no relevant financial or non-financial interests to disclose. Authors' contributions Study concepts: NBC, HY Study design: NBC, HY, EY Data acquisition: NBC, MU Quality control of data and algorithms: HY, SİO, MKP Data analysis and interpretation: EY, SİO, MKP Statistical analysis: HY, EY Manuscript preparation: NBC Manuscript editing: KT, HY Manuscript review: NBC, KT, HY All authors have read and approved the manuscript. Funding No financial funding was received. Availability of data and materials All of the material is owned by the authors and/or no permissions are required. References Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394-424. Schulz GB, Grimm T, Buchner A, Jokisch F, Kretschmer A, Casuscelli J, et al. Surgical High-risk Patients With ASA ≥ 3 Undergoing Radical Cystectomy: Morbidity, Mortality, and Predictors for Major Complications in a High-volume Tertiary Center. Clin Genitourin Cancer. 2018;16(6):e1141-e9. Cumberbatch MGK, Jubber I, Black PC, Esperto F, Figueroa JD, Kamat AM, et al. Epidemiology of Bladder Cancer: A Systematic Review and Contemporary Update of Risk Factors in 2018. Eur Urol. 2018;74(6):784-95. Coussens LM, Werb Z. Inflammation and cancer. Nature. 2002;420(6917):860-7. Lucca I, Jichlinski P, Shariat SF, Rouprêt M, Rieken M, Kluth LA, et al. The Neutrophil-to-lymphocyte Ratio as a Prognostic Factor for Patients with Urothelial Carcinoma of the Bladder Following Radical Cystectomy: Validation and Meta-analysis. Eur Urol Focus. 2016;2(1):79-85. Schuettfort VM, D'Andrea D, Quhal F, Mostafaei H, Laukhtina E, Mori K, et al. A panel of systemic inflammatory response biomarkers for outcome prediction in patients treated with radical cystectomy for urothelial carcinoma. BJU Int. 2022;129(2):182-93. Qi Q, Zhuang L, Shen Y, Geng Y, Yu S, Chen H, et al. A novel systemic inflammation response index (SIRI) for predicting the survival of patients with pancreatic cancer after chemotherapy. Cancer. 2016;122(14):2158-67. Grossmann NC, Schuettfort VM, Pradere B, Rajwa P, Quhal F, Mostafaei H, et al. Impact of preoperative systemic immune-inflammation Index on oncologic outcomes in bladder cancer patients treated with radical cystectomy. Urol Oncol. 2022;40(3):106.e11-.e19. Ojerholm E, Smith A, Hwang WT, Baumann BC, Tucker KN, Lerner SP, et al. Neutrophil-to-lymphocyte ratio as a bladder cancer biomarker: Assessing prognostic and predictive value in SWOG 8710. Cancer. 2017;123(5):794-801. Napolitano L, Barone B, Reccia P, De Luca L, Morra S, Turco C, et al. Preoperative monocyte-to-lymphocyte ratio as a potential predictor of bladder cancer. J Basic Clin Physiol Pharmacol. 2022;33(6):751-7. Wang X, Ni X, Tang G. Prognostic Role of Platelet-to-Lymphocyte Ratio in Patients With Bladder Cancer: A Meta-Analysis. Front Oncol. 2019;9:757. Erdem GK, Omurca SI, Cakir EB, Bayrak BY. Prediction of pathological grade in prostate cancer: an ensemble deep learning-based whole slide image classification model. The European Physical Journal Special Topics. 2025:1-13. Breiman L. Random forests Mach Learn 45 (1): 5–32. ed; 2001. Friedman JH. Greedy function approximation: a gradient boosting machine. Annals of statistics. 2001:1189-232. Ke G, Meng Q, Finley T, Wang T, Chen W, Ma W, et al. Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems. 2017;30. Prokhorenkova L, Gusev G, Vorobev A, Dorogush AV, Gulin A. CatBoost: unbiased boosting with categorical features. Advances in neural information processing systems. 2018;31. Mangalathu S, Hwang S-H, Jeon J-S. Failure mode and effects analysis of RC members based on machine-learning-based SHapley Additive exPlanations (SHAP) approach. Engineering Structures. 2020;219:110927. Obermeyer Z, Emanuel EJ. Predicting the Future - Big Data, Machine Learning, and Clinical Medicine. N Engl J Med. 2016;375(13):1216-9. Yilmaz H, Cinar NB, Avci IE, Telli E, Uslubas AK, Teke K, et al. The systemic inflammation response index: An independent predictive factor for survival outcomes of bladder cancer stronger than other inflammatory markers. Urol Oncol. 2023;41(5):256.e1-.e8. Shakhssalim N, Talebi A, Pahlevan-Fallahy MT, Sotoodeh K, Alavimajd H, Borumandnia N, et al. Lifestyle and occupational risks assessment of bladder cancer using machine learning-based prediction models. Cancer Rep (Hoboken). 2023;6(9):e1860. Wu E, Hadjiiski LM, Samala RK, Chan HP, Cha KH, Richter C, et al. Deep Learning Approach for Assessment of Bladder Cancer Treatment Response. Tomography. 2019;5(1):201-8. Zhao CC, Bjurlin MA, Wysock JS, Taneja SS, Huang WC, Fenyo D, et al. Machine learning decision support model for radical cystectomy discharge planning. Urol Oncol. 2022;40(10):453.e9-.e18. Xu X, Zhang X, Tian Q, Zhang G, Liu Y, Cui G, et al. Three-dimensional texture features from intensity and high-order derivative maps for the discrimination between bladder tumors and wall tissues via MRI. Int J Comput Assist Radiol Surg. 2017;12(4):645-56. Brieu N, Gavriel CG, Nearchou IP, Harrison DJ, Schmidt G, Caie PD. Automated tumour budding quantification by machine learning augments TNM staging in muscle-invasive bladder cancer prognosis. Sci Rep. 2019;9(1):5174. Galsky MD, Stensland K, Sfakianos JP, Mehrazin R, Diefenbach M, Mohamed N, et al. Comparative Effectiveness of Treatment Strategies for Bladder Cancer With Clinical Evidence of Regional Lymph Node Involvement. J Clin Oncol. 2016;34(22):2627-35. Ji J, Zhang T, Zhu L, Yao Y, Mei J, Sun L, et al. Using machine learning to develop preoperative model for lymph node metastasis in patients with bladder urothelial carcinoma. BMC Cancer. 2024;24(1):725. Mathieu R, Lucca I, Rouprêt M, Briganti A, Shariat SF. The prognostic role of lymphovascular invasion in urothelial carcinoma of the bladder. Nat Rev Urol. 2016;13(8):471-9. Tian YF, Zhou H, Yu G, Wang J, Li H, Xia D, et al. Prognostic significance of lymphovascular invasion in bladder cancer after surgical resection: A meta-analysis. J Huazhong Univ Sci Technolog Med Sci. 2015;35(5):646-55. Garg T, Chen LY, Kim PH, Zhao PT, Herr HW, Donat SM. Preoperative serum albumin is associated with mortality and complications after radical cystectomy. BJU Int. 2014;113(6):918-23. Tables Table 1. Demographic and Clinical Data No. Patients. n 241 Age (y); median (IQR) 65 (60-70) Gender Male; n (%) 205 (85.1) Female. n (%) 36 (14.9) Hydronephrosis. n (%) 82 (34) e-GFR before Cystectomy (ml/min/1.73 m 2 ), median (IQR) 78 (52.5-91) ACCI. median (IQR) 5 (4-6) Clinical Stage cT1. n (%) 17 (7.1) cT2. n (%) 118 (49) cT3. n (%) 74 (30.7) cT4. n (%) 32 (13.3) Neoadjuvant chemotherapy. n (%) 41 (17.0) Cystectomy characteristics pT stage pT0. n (%) 14 (5.8) pTa. n (%) 6 (2.5) pT1. n (%) 24 (10) pT2. n (%) 58 (24.1) pT3. n (%) 75 (31.1) pT4. n (%) 62 (25.7) pN stage pN0. n (%) 151 (62.7) pN1. n (%) 31 (12.9) pN2. n (%) 58 (24.1) PN3. n (%) 1 (0.4) Variant histology. n (%) 55 (22.8) Lymphovascular invasion. n (%) 105 (43.6) Positive surgical margin. n (%) 15 (6.2) Necrosis. n (%) 68 (28.2) Adjuvant chemotherapy. n (%) 42 (17.4) Tur-Cystectomy interval time (mo), median (IQR) 2.1 (1.3-4.4) Intraoperative blood loss (ml), median (IQR) 1100 (100-6000) Operative time (min), median (IQR) 420 (300-690) Length of stay (d), median (IQR) 12 (7-56) Peroperative Transfusion, n (%) 216 (89.6) RS Pathology Cis, n (%) 66 (27.4) Adenocarcinoma of the prostate. n (%) 60 (24.9) Death, n (%) 145 (60.2) Preoperative blood parameters Hemoglobin, median (IQR) 12 (10.8-13.3) Creatinine, median (IQR) 1.2 (0.8-1.3) Albumin, median (IQR) 3.9 (3.4-4.2) Total Protein, median (IQR) 7 (6.5-7.4) Neutrophils, median (IQR) 5.3 (3.9-7.3) Lymphocytes, median (IQR) 1.8 (1.3-2.3) Monocytes, median (IQR) 0.61 (0.47-0.79) Platelet, median (IQR) 235 (191.5-307.5) Inflammatory parameters NLR, median (IQR) 2.8 (2.1-4.3) MLR, median (IQR) 0.34 (0.25-0.47) PLR, median (IQR) 135.2(101.7-185) SII, median (IQR) 715 (457-1108.2) SIRI, median (IQR) 1.7 (1.1-3) Abbreviations: e-GFR. estimated glomerular filtration rate; ACCI. age adjusted charlson comorbidity index; IQR. inter quartile rangeLVI, lymphovascular invasion; SIRI, systemic inflammatory response index; NLR, Neutrophil to lymphocyte ratio; MLR, Monocyte to lymphocyte ratio; PLR, Platelet to lymphocyte ratio; SII, Systemic inflammatory index Table 2. Performance analysis of DS1 (30 original features) and DS-2 (original features plus all inflammatory markers) on experimental results of various models by evaluation metrics DS-1 DS-2 Model F1-Score Precision Recall Accuracy ROC_AUC F1-Score Precision Recall Accuracy ROC_AUC CART 0,728565 0,751077 0,719047 0,6846 0,687394 0,731773 0,766374 0,714761 0,693 0,701518 CatBoost 0,771186 0,744684 0,807142 0,713166 0,771253 0,770551 0,76014 0,784761 0,7175 0,747312 GBM 0,780757 0,754495 0,813333 0,726166 0,769201 0,782028 0,734484 0,840952 0,7176 0,757751 LightGBM 0,755676 0,768184 0,750476 0,7095 0,71228 0,758028 0,763158 0,758095 0,7096 0,719285 RandomForests 0,784117 0,757227 0,820476 0,7298 0,75773 0,782969 0,725811 0,854285 0,7178 0,766013 XGBoost 0,785037 0,76984 0,807142 0,7341 0,748216 0,774986 0,743152 0,813809 0,7178 0,737179 Logistic Regression 0,7464969 0,742808 0,758095 0,6963 0,745444 0,762872 0,751794 0,778571 0,713 0,74238 SVM 0,738954 0,750263 0,736666 0,6878 0,745925 0,746838 0,757916 0,743333 0,7045 0,737074 MLP 0,773728 0,762207 0,792857 0,7255 0,77301 0,762778 0,767999 0,764761 0,7171 0,750465 Table.3: Changes with stepwise addition of inflammatory markers (SIRI, LMR, NLR, PLR, SII) to the original dataset on the evaluation of F1 Scores in a Random Forest model using DS-3. Combinations F1-Score Precision Recall Accuracy ROC_AUC DS1+SIRI 0.791041 0.750858 0.840476 0.7341 0.786629 DS1+SIRI+LMR 0.788862 0.747563 0.841428 0.7301 0.775703 DS1+SIRI+NLR 0.793419 0.760625 0.835238 0.7385 0.779417 DS1+SIRI+NLR+PLR 0.801126 0.737930 0.883333 0.7344 0.781582 DS1+SIRI+NLR+PLR+SII 0.789543 0.745874 0.841428 0.7301 0.754052 Additional Declarations No competing interests reported. Supplementary Files SuppFig.jpg Supplementary Figure 1: The following combinations were sequentially created for the Random Forests model (DS-3): Original + SIRI, Original + SIRI + LMR, Original + SIRI + NLR, Original + SIRI + NLR + PLR, and Original + SIRI + NLR + PLR + SII. Among these sub-variants, the highest F1 score of 0.80 was achieved by sequentially adding SIRI, NLR, and PLR (Table 3). Cite Share Download PDF Status: Published Journal Publication published 21 Oct, 2025 Read the published version in World Journal of Urology → Version 1 posted Editorial decision: Revision requested 18 Jul, 2025 Reviews received at journal 08 Jul, 2025 Reviews received at journal 08 Jul, 2025 Reviews received at journal 07 Jul, 2025 Reviews received at journal 04 Jul, 2025 Reviewers agreed at journal 01 Jul, 2025 Reviewers agreed at journal 30 Jun, 2025 Reviewers agreed at journal 30 Jun, 2025 Reviewers agreed at journal 11 Jun, 2025 Reviewers invited by journal 29 May, 2025 Editor assigned by journal 28 May, 2025 Submission checks completed at journal 28 May, 2025 First submitted to journal 21 May, 2025 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-6717746","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":464629686,"identity":"3822dc39-62c1-43a4-8ac5-7c2ff22da9aa","order_by":0,"name":"NACİ BURAK ÇINAR","email":"","orcid":"","institution":"Department of Urology, Kutahya City Hospital, Kutahya","correspondingAuthor":false,"prefix":"","firstName":"NACİ","middleName":"BURAK","lastName":"ÇINAR","suffix":""},{"id":464629687,"identity":"ab079f3e-8796-401c-bc24-63fafaa9fcf0","order_by":1,"name":"HASAN YILMAZ","email":"data:image/png;base64,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","orcid":"","institution":"Department of Urology, Kocaeli University School of Medicine, Kocaeli","correspondingAuthor":true,"prefix":"","firstName":"HASAN","middleName":"","lastName":"YILMAZ","suffix":""},{"id":464629688,"identity":"e0db7f77-8a4b-41b8-bdcd-316d976291ee","order_by":2,"name":"EFE YILMAZ TAŞYÜREK","email":"","orcid":"","institution":"Faculty of Engineering, Department of Computer Engineering, Kocaeli University, Kocaeli","correspondingAuthor":false,"prefix":"","firstName":"EFE","middleName":"YILMAZ","lastName":"TAŞYÜREK","suffix":""},{"id":464629689,"identity":"1aafee45-98eb-4cef-9e9d-675ec075a8b7","order_by":3,"name":"MELTEM KURT PEHLIVANOĞLU","email":"","orcid":"","institution":"Faculty of Engineering, Department of Computer Engineering, Kocaeli University, Kocaeli","correspondingAuthor":false,"prefix":"","firstName":"MELTEM","middleName":"KURT","lastName":"PEHLIVANOĞLU","suffix":""},{"id":464629690,"identity":"9e9c3f0f-9c55-4217-ba77-85d624ec67cf","order_by":4,"name":"SEVINÇ İLHAN OMURCA","email":"","orcid":"","institution":"Faculty of Engineering, Department of Computer Engineering, Kocaeli University, Kocaeli","correspondingAuthor":false,"prefix":"","firstName":"SEVINÇ","middleName":"İLHAN","lastName":"OMURCA","suffix":""},{"id":464629691,"identity":"d4634d8d-14c4-4aba-847f-dc3d6d073ed3","order_by":5,"name":"MUHLİS ÜNAL","email":"","orcid":"","institution":"Faculty of Engineering, Department of Computer Engineering, Kocaeli University, Kocaeli","correspondingAuthor":false,"prefix":"","firstName":"MUHLİS","middleName":"","lastName":"ÜNAL","suffix":""},{"id":464629692,"identity":"3da79e96-a8e8-40c4-a60a-7961d326a7ca","order_by":6,"name":"KEREM TEKE","email":"","orcid":"","institution":"Department of Urology, Kocaeli University School of Medicine, Kocaeli","correspondingAuthor":false,"prefix":"","firstName":"KEREM","middleName":"","lastName":"TEKE","suffix":""}],"badges":[],"createdAt":"2025-05-21 15:08:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6717746/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6717746/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00345-025-05978-7","type":"published","date":"2025-10-21T16:16:39+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":83837501,"identity":"28cc035a-6845-4f72-8c89-04294e1664af","added_by":"auto","created_at":"2025-06-03 13:24:05","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":64618,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure 1-A.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSHAP values of DS-1, which contains 29 original features, based on evaluation metrics showing the contribution and importance of individual features to the predictive performance of the Random Forest model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFigure 1-B.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSHAP values of DS-2, which contains the original 29 features plus all inflammatory markers in no particular order, based on evaluation metrics showing the contribution and importance of individual features to the predictive performance of the Random Forest model\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFigure 1-C.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSHAP values of F1 Scores in Random Forest model are shown. Changes with stepwise addition of the inflammatory markers, SIRI, LMR, NLR, PLR, and SII to create DS-3, which in addition contained the 29 original features, based on evaluation metrics showing the contribution and importance of individual features to the predictive performance of the Random Forest model.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6717746/v1/7e7274333f3139b677026da1.png"},{"id":94490323,"identity":"06cdd181-091c-42ab-b0e6-eaca9bd98f33","added_by":"auto","created_at":"2025-10-27 17:09:09","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1075084,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6717746/v1/4e936da8-7791-464e-8f49-b289b1188195.pdf"},{"id":83837497,"identity":"8cba742c-f2ce-4139-99a4-4f6c24bb60ba","added_by":"auto","created_at":"2025-06-03 13:24:03","extension":"jpg","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":69526,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Figure 1:\u003c/strong\u003e The following combinations were sequentially created for the Random Forests model (DS-3): Original + SIRI, Original + SIRI + LMR, Original + SIRI + NLR, Original + SIRI + NLR + PLR, and Original + SIRI + NLR + PLR + SII. Among these sub-variants, the highest F1 score of 0.80 was achieved by sequentially adding SIRI, NLR, and PLR (Table 3).\u003c/p\u003e","description":"","filename":"SuppFig.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6717746/v1/8e5bf164eb03d97bf957c5e9.jpg"}],"financialInterests":"No competing interests reported.","formattedTitle":"Evaluation of Inflammatory Markers in Survival Analysis of Patients Undergoing Radical Cystectomy Using Machine Learning","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBladder cancer is the 11th most common cancer worldwide, with approximately 550,000 new cases diagnosed annually(1), and it is a significant cause of morbidity and mortality(2). Muscle-invasive bladder cancer (MIBC) accounts for 25% of cases, with radical cystectomy (RC) and pelvic lymph node dissection as the standard treatment. However, despite the radical treatment modalities, the 5-year overall survival rate remains at 60%(3). Thus, to determine the optimal treatment strategy for improvement of patient survival rates, predictive models may be beneficial. There are existing models for prediction of post-cystectomy survival but an ideal model for this purpose has not been established in the literature, yet.\u003c/p\u003e \u003cp\u003eMachine learning (ML) has emerged as a powerful tool in medicine, particularly in oncology, where it facilitates survival prediction and risk stratification. ML models have demonstrated significant advances in predictive models for outcomes. In the context of survival after RC, ML algorithms can simply analyse a wide range of patient data, including demographic data, clinical and laboratory results, and imaging findings. In addition, ML algorithms may also identify predictive effects of novel markers on patient survival. Thus, ML algorithms can further contribute to specialists in pre-operative risk management and the future course of the disease. Moreover, ML algorithms have the potential to become sustainable by adapting to new patient data.\u003c/p\u003e \u003cp\u003eInflammation plays an important role in tumor development and progression (4). Systemic inflammatory response cytokines are induced by the tumor microenvironment and this has led to the use of inflammatory markers as clinical and pathological predictive and prognostic factors in many studies(5). Numerous markers were found to be predictive for bladder cancer in the literature. However, a ML-based study concluded that the systemic inflammatory response (SIR) biomarker panel did not significantly improve the predictive ability of oncological outcomes in patients undergoing RC (6). In the present study, the aim was to create novel ML models using patient demographic, clinical and pathological data in patients undergoing RC. A secondary aim was to evaluate whether inflammatory markers contributed to these models.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003eThis study included consecutive patients who underwent open RC, bilateral pelvic lymph node dissection, and urinary diversion (ileal loop or Studer pouch) for bladder cancer between 2009 and 2021.\u003c/p\u003e \u003cp\u003eDemographic and clinic data of patients including preoperative, intraoperative, and postoperative information were analyzed using 30 original features (target feature included) in ML models (DS-1). Oncological outcomes for this study included time to death (overall survival). Preoperative blood parameters were obtained one day before the RC procedure. Numerous inflammatory markers were derived from these parameters. These markers included the systemic inflammatory response index (SIRI) (7), systemic immuno-inflammation index (SII)(8), neutrophil-to-lymphocyte ratio (NLR) (9), monocyte-to-lymphocyte ratio (MLR) (10), and platelet-to-lymphocyte ratio (PLR)(11). To more clearly evaluate the contribution of inflammatory markers, the DS-1 dataset was systematically expanded by adding inflammatory markers to create the DS-2 dataset. The predictive effectiveness of five derived inflammatory markers was evaluated based on the outcome of the death feature.\u003c/p\u003e \u003cp\u003eTo better assess the impact of inflammatory markers, we constructed several data subsets to explain the mortality feature using five specific inflammatory markers. We then recorded feature importance rankings derived from leading machine learning models, including CatBoost, XGBoost, and Random Forest. Based on these rankings, we systematically expanded the original dataset by sequentially incorporating markers according to their relative importance. The marker with the highest importance was initially added to the baseline feature set. Subsequently, the remaining markers were introduced individually, in descending order of importance. If the inclusion of a marker did not enhance the F1 score, it was excluded from the cumulative addition process. Ultimately, the highest performance scores were achieved using the Random Forest model. This approach aimed to evaluate the stepwise effect of each marker and determine the optimal combination of features for survival prediction, resulting in the DS-3 dataset. Supplementary Fig.\u0026nbsp;1. illustrates the feature importance ranking generated by the Random Forest model for the subset designed to explain the mortality feature using the five inflammatory markers\u003c/p\u003e \u003cp\u003eThe performance of applied ML algorithms was evaluated using F1-Score, Precision, Recall, Accuracy, and area under the curve-receiver operator characteristics (AUC-ROC) curve metrics(12).\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eMachine Learning Algorithms\u003c/h2\u003e \u003cp\u003eThis section summarizes the classification models and the Shapley Additive Explanations (SHAP) method used for feature selection. Supervised ML models, including Support Vector Machines (SVM), Multi-Layer Perceptron (MLP), Logistic Regression, Decision Trees, and ensemble learning models such as Random Forest were used(13). In addition, Gradient Boosting Machines (GBM), eXtreme Gradient Boosting (XGBoost)(14), Light Gradient Boosting Machine (LightGBM)(15), and Category Boosting (CatBoost) (16) were also applied. The rationale for the selection of these elements was as follows. Logistic Regression models linear relationships, while Decision Trees simplifies complex datasets into decision rules. Random Forest combines multiple decision trees, and Gradient Boosting algorithms (XGBoost, LightGBM, and CatBoost) enhance classification performance by combining weak learners. XGBoost excels in scalability, CatBoost optimizes categorical features with faster convergence, and LightGBM processes large datasets efficiently by growing trees vertically. SVM separates data using hyperplanes, and MLP is a neural network foundational to deep learning.\u003c/p\u003e \u003cp\u003eThe sensitivity of classification models to dependent variables can be investigated using SHAP values (17). The SHAP method is capable of assigning a degree of importance for each of the input parameters separately. In the present study, SHAP values were determined to examine the importance of the current clinical features and the importance of additional inflammatory markers.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eEvaluation Metrics\u003c/h3\u003e\n\u003cp\u003eModel performance was evaluated across five metrics: Accuracy, Precision, Recall, F1-score, and ROC-AUC. Considering the class imbalance, the primary performance metric was selected as the F1-score, which balances Precision and Recall metrics. Grid Search-CV and K-Fold Cross-Validation methods were employed to determine the optimal hyperparameters of the models and enhance their overall performance. During the hyperparameter optimization process, various parameter combinations were systematically evaluated, and each combination was tested using 10-fold cross-validation. This approach enabled the identification of the parameters that yielded the highest performance for each model. Once the optimal parameters were determined, a final 10-fold cross-validation was conducted on the entire dataset to obtain the ultimate performance metrics. For this, the dataset was randomly divided into ten equal parts, and each model was trained on nine of these subsets while being tested on the remaining one. This process was repeated ten times, ensuring each subset was a test set once. Applying this technique, the model underwent a comprehensive evaluation across the entire dataset. This primary objective of this approach was to ensure that the models performed well on a specific patient group and to guarantee a generalizable performance, making the models robust across different data distributions.\u003c/p\u003e\n\u003ch3\u003eStatistical analyses\u003c/h3\u003e\n\u003cp\u003eThe Kolmogorov-Smirnov test was used to determine the normality of the numeric data. Categorical variables are expressed as count/frequency (n) and percentage (%), and continuous variables are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD) or, if not normally distributed, median and interquartile range (IQR). The Kaplan-Meier method was used for OS calculations, using SPSS, version 21 (IBM Inc., Armonk, NY, USA)\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eA total of 251 patients undergoing RC were identified over the study period. Patients who had previously received radiotherapy (n\u0026thinsp;=\u0026thinsp;10) were excluded, resulting in a total of 241 patients. The median age was 65 years. Approximately half of the patients presented with advanced clinical and pathological T stage disease (T3 or T4). The median (IQR) follow-up duration was 20 (9\u0026ndash;52) months. The mortality rate was 60.2% (145/241). Two and 5-year overall survival (OS) rates were 54.7% and 37.2%, respectively. The median OS was 30.00 months (95% CI 15.06\u0026ndash;44.93). The demographic and clinical characteristics of the patients are presented in Table\u0026nbsp;1.\u003c/p\u003e\n\u003cp\u003eThe performance of nine different ML models was thoroughly evaluated over DS-1 (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). These analyses resulted in an F1 score between 0.72\u0026ndash;0.78. Random Forests and XGBoost models achieved the highest score of 0.78. All inflammatory markers were cumulatively added to the DS-1 to create DS-2, however no significant improvement was observed in F1 scores, ranging from 0.73 to 0.78 on subsequent performance analysis.\u003c/p\u003e\n\u003cp\u003eTo investigate the effect of inflammatory markers more deeply, the DS-1 dataset was systematically expanded by adding inflammatory markers in an order based on the feature importance scores obtained from the prominent models CatBoost, XGBoost, and RandomForests algorithms to create DS-3. Initially, the most predictive inflammatory marker was added to the original feature set alone. Then, other derived markers were added in decreasing order according to their predictive value. While following this sequence, if a specific inflammatory marker did not increase the F1-score metric, it was removed from this cumulatively advanced addition. The sequence was: Original\u0026thinsp;+\u0026thinsp;SIRI, Original\u0026thinsp;+\u0026thinsp;SIRI\u0026thinsp;+\u0026thinsp;LMR, Original\u0026thinsp;+\u0026thinsp;SIRI\u0026thinsp;+\u0026thinsp;NLR, Original\u0026thinsp;+\u0026thinsp;SIRI\u0026thinsp;+\u0026thinsp;NLR\u0026thinsp;+\u0026thinsp;PLR, and Original\u0026thinsp;+\u0026thinsp;SIRI\u0026thinsp;+\u0026thinsp;NLR\u0026thinsp;+\u0026thinsp;PLR\u0026thinsp;+\u0026thinsp;SII. In these subvariants, sequential addition of SIRI, NLR and PLR to the original features in the Random Forest achieved the highest F1 score of 0.80 \u003cstrong\u003e(Table\u0026nbsp;3).\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e-A, Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e-B, and Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e-C present the explainability analysis of the Random Forest model using SHAP values on different datasets. These figures are designed to assess the impact of various features on the model\u0026apos;s performance in predicting survival. In the SHAP analyses of DS-1, T stage, preoperative albumin, and lympho-vascular invasion (LVI) were identified as the most influential factors in model classification performance (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eA). In the DS-2, again T stage, preoperative albumin, and LVI were identified as the most influential factors, while SIRI and LMR were observed as the most effective markers among derived markers (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eB). In DS-3, T stage, preoperative albumin, and preoperative platelet count were observed as the most effective markers seen as the most important factors, and the next most important was LVI (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eC). Moreover, the SIRI, PLR, and NLR variables had positive effects on the performance of the model, with NLR and SIRI ranked especially highly.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eCurrently, ML has increasingly contributed to the diagnosis and treatment of various diseases (18). Bladder tumors have increasingly been included in these studies In the literature review for the present research, there were a few studies related to bladder cancer and ML. Schuettfort et al analysed predictive factors in patients who underwent RC (6). Notably, their study which was the first ML-based study assessing the prognostic value of inflammatory markers focused specifically on the systemic inflammatory response (SIR) biomarker panel. This panel included the albumin-globulin ratio, NLR, De Ritis ratio, MLR, and the modified Glasgow prognostic score. The authors concluded that the created models including SIR biomarkers did not significantly improve the predictive ability of oncological outcomes(6). In the current study we found important differences to the study of Schuettfort et al. First, we comprehensively evaluated the performance of nine different ML models. Second, we added the inflammatory markers to the ML models in a specific sequence starting with the one with the highest contribution. Like the previous study, including all derived markers in a cumulative fashion did not improve the F1 scores. However, integrating the inflammatory markers with the most effective first to original data set, that is sequential addition of SIRI, NLR and PLR, achieved the highest F1 score of al combinations. This ML model was the most reliable model for survival prediction in our cohort. Of note, the SII and LMR markers made no contribution to model improvementThis is the first published study to demonstrate a significant improvement in performance of ML models with inflammatory marker data in the prediction of bladder cancer survival after RC. Furthermore, our findings align with our previous study based on the same dataset(19). In that study, multivariate Cox regression analyses demonstrated that both categorical and continuous SIRI were independent predictors of recurrence-free survival and overall survival.\u003c/p\u003e \u003cp\u003eIn another study, ML strategies were employed to assess the influence of lifestyle and occupational risk factors on bladder cancer(20). Wu et al. identified significant results using different deep convolutional neural network (DCNN) models to predict the response of bladder lesions to chemotherapy(21). Furthermore, an ML-based decision support model was developed in a large cohort to enhance the reliable and effective management of patients at high risk for postoperative complications following cystectomy and to optimize discharge planning (22). Tsai et al. analyzed five models incorporating calcium, alkaline phosphatase, albumin, urine ketones, occult hematuria, and liver transaminase data to differentiate bladder cancer from renal cancer, cystitis, and prostate cancer. Furthermore, ML has demonstrated potential in the non-invasive staging of bladder cancer through 3D texture analysis of bladder wall CT/MRI images, particularly in cases with challenging diagnostic features in differentiating malignant areas from normal urothelium during cystoscopy (23). Brieu et al. used a Random Forest model to accurately quantify tumor budding in muscle-invasive bladder cancer patients through immunofluorescence imaging, presenting it as a prognostic tool which appeared superior to TNM staging (24). In contrast, Galsky et al evaluated the effectiveness of treatment strategies for bladder cancer with clinical evidence of regional lymph node involvement (25). Then, ML models were developed to preoperatively predict lymph node metastasis, a known poor prognostic factor for survival with a 14-variable SVM model achieving the best performance(26).\u003c/p\u003e \u003cp\u003eIn the present study, SHAP analysis identified T stage, preoperative albumin, preoperative platelet count and the presence of LVI as having the most contribution to model prediction power in the Random Forest model. The SHAP analysis of three different models revealed similar key features. The results we obtained are consistent with previous studies in the literature (27\u0026ndash;29). The prominence of NLR and SIRI in the most reliable model we developed shows that the derived markers not only make a significant contribution to our model but should also be considered in other modelling exercises for prognostic markers in bladder and other cancers.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eLimitation\u003c/h2\u003e \u003cp\u003eThis study has some important limitations. First, the inclusion of a large number of features derived from patient data, coupled with a limited sample size, presents challenges in developing a robust ML model. Nevertheless, we anticipate that as the patient cohort expands, the predictive power of our model will improve, thereby yielding more reliable results. Furthermore, the retrospective design of the study constitutes another limitation. We were not able to use cancer-specific survival as an outcome variable because of insufficient data. Using overall survival instead may have affected our results.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eWe evaluated survival prediction in patients with bladder tumors using nine different ML models. We found that T stage, preoperative albumin and LVI contributed the most to accuracy of these models. In addition, adding the inflammatory markers SIRI, NLR and PLR to demographic data resulted in the highest F1 score. These findings demonstrate that inflammatory markers can contribute positively to ML models in this disease. However, more studies with larger patient numbers are needed to increase the predictive power of such ML models.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical Approval:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis retrospective study was approved by the local ethics committee of Kocaeli University (Approval number: K\u0026Uuml; GOKAEK-2025/148). The study was conducted in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to Participate:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWritten informed consent was obtained from all patients prior to surgery.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for Publication:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors did not receive support from any organization for the submitted work. The authors have no relevant financial or non-financial interests to disclose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStudy concepts:\u003c/strong\u003e NBC, HY\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStudy design:\u003c/strong\u003e NBC, HY, EY\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData acquisition:\u003c/strong\u003e NBC, MU\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eQuality control of data and algorithms:\u003c/strong\u003e HY, SİO, MKP\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData analysis and interpretation:\u003c/strong\u003e EY, SİO, MKP\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis:\u003c/strong\u003e HY, EY\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eManuscript preparation:\u003c/strong\u003e NBC\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eManuscript editing:\u003c/strong\u003e KT, HY\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eManuscript review:\u003c/strong\u003e NBC, KT, HY\u003c/p\u003e\n\u003cp\u003eAll authors have read and approved the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNo financial funding was received.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll of the material is owned by the authors and/or no permissions are required.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394-424.\u003c/li\u003e\n\u003cli\u003eSchulz GB, Grimm T, Buchner A, Jokisch F, Kretschmer A, Casuscelli J, et al. Surgical High-risk Patients With ASA \u0026ge; 3 Undergoing Radical Cystectomy: Morbidity, Mortality, and Predictors for Major Complications in a High-volume Tertiary Center. Clin Genitourin Cancer. 2018;16(6):e1141-e9.\u003c/li\u003e\n\u003cli\u003eCumberbatch MGK, Jubber I, Black PC, Esperto F, Figueroa JD, Kamat AM, et al. Epidemiology of Bladder Cancer: A Systematic Review and Contemporary Update of Risk Factors in 2018. Eur Urol. 2018;74(6):784-95.\u003c/li\u003e\n\u003cli\u003eCoussens LM, Werb Z. Inflammation and cancer. Nature. 2002;420(6917):860-7.\u003c/li\u003e\n\u003cli\u003eLucca I, Jichlinski P, Shariat SF, Roupr\u0026ecirc;t M, Rieken M, Kluth LA, et al. The Neutrophil-to-lymphocyte Ratio as a Prognostic Factor for Patients with Urothelial Carcinoma of the Bladder Following Radical Cystectomy: Validation and Meta-analysis. Eur Urol Focus. 2016;2(1):79-85.\u003c/li\u003e\n\u003cli\u003eSchuettfort VM, D\u0026apos;Andrea D, Quhal F, Mostafaei H, Laukhtina E, Mori K, et al. A panel of systemic inflammatory response biomarkers for outcome prediction in patients treated with radical cystectomy for urothelial carcinoma. BJU Int. 2022;129(2):182-93.\u003c/li\u003e\n\u003cli\u003eQi Q, Zhuang L, Shen Y, Geng Y, Yu S, Chen H, et al. A novel systemic inflammation response index (SIRI) for predicting the survival of patients with pancreatic cancer after chemotherapy. Cancer. 2016;122(14):2158-67.\u003c/li\u003e\n\u003cli\u003eGrossmann NC, Schuettfort VM, Pradere B, Rajwa P, Quhal F, Mostafaei H, et al. Impact of preoperative systemic immune-inflammation Index on oncologic outcomes in bladder cancer patients treated with radical cystectomy. Urol Oncol. 2022;40(3):106.e11-.e19.\u003c/li\u003e\n\u003cli\u003eOjerholm E, Smith A, Hwang WT, Baumann BC, Tucker KN, Lerner SP, et al. Neutrophil-to-lymphocyte ratio as a bladder cancer biomarker: Assessing prognostic and predictive value in SWOG 8710. Cancer. 2017;123(5):794-801.\u003c/li\u003e\n\u003cli\u003eNapolitano L, Barone B, Reccia P, De Luca L, Morra S, Turco C, et al. Preoperative monocyte-to-lymphocyte ratio as a potential predictor of bladder cancer. J Basic Clin Physiol Pharmacol. 2022;33(6):751-7.\u003c/li\u003e\n\u003cli\u003eWang X, Ni X, Tang G. Prognostic Role of Platelet-to-Lymphocyte Ratio in Patients With Bladder Cancer: A Meta-Analysis. Front Oncol. 2019;9:757.\u003c/li\u003e\n\u003cli\u003eErdem GK, Omurca SI, Cakir EB, Bayrak BY. Prediction of pathological grade in prostate cancer: an ensemble deep learning-based whole slide image classification model. The European Physical Journal Special Topics. 2025:1-13.\u003c/li\u003e\n\u003cli\u003eBreiman L. Random forests Mach Learn 45 (1): 5\u0026ndash;32. ed; 2001.\u003c/li\u003e\n\u003cli\u003eFriedman JH. Greedy function approximation: a gradient boosting machine. Annals of statistics. 2001:1189-232.\u003c/li\u003e\n\u003cli\u003eKe G, Meng Q, Finley T, Wang T, Chen W, Ma W, et al. Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems. 2017;30.\u003c/li\u003e\n\u003cli\u003eProkhorenkova L, Gusev G, Vorobev A, Dorogush AV, Gulin A. CatBoost: unbiased boosting with categorical features. Advances in neural information processing systems. 2018;31.\u003c/li\u003e\n\u003cli\u003eMangalathu S, Hwang S-H, Jeon J-S. Failure mode and effects analysis of RC members based on machine-learning-based SHapley Additive exPlanations (SHAP) approach. Engineering Structures. 2020;219:110927.\u003c/li\u003e\n\u003cli\u003eObermeyer Z, Emanuel EJ. Predicting the Future - Big Data, Machine Learning, and Clinical Medicine. N Engl J Med. 2016;375(13):1216-9.\u003c/li\u003e\n\u003cli\u003eYilmaz H, Cinar NB, Avci IE, Telli E, Uslubas AK, Teke K, et al. The systemic inflammation response index: An independent predictive factor for survival outcomes of bladder cancer stronger than other inflammatory markers. Urol Oncol. 2023;41(5):256.e1-.e8.\u003c/li\u003e\n\u003cli\u003eShakhssalim N, Talebi A, Pahlevan-Fallahy MT, Sotoodeh K, Alavimajd H, Borumandnia N, et al. Lifestyle and occupational risks assessment of bladder cancer using machine learning-based prediction models. Cancer Rep (Hoboken). 2023;6(9):e1860.\u003c/li\u003e\n\u003cli\u003eWu E, Hadjiiski LM, Samala RK, Chan HP, Cha KH, Richter C, et al. Deep Learning Approach for Assessment of Bladder Cancer Treatment Response. Tomography. 2019;5(1):201-8.\u003c/li\u003e\n\u003cli\u003eZhao CC, Bjurlin MA, Wysock JS, Taneja SS, Huang WC, Fenyo D, et al. Machine learning decision support model for radical cystectomy discharge planning. Urol Oncol. 2022;40(10):453.e9-.e18.\u003c/li\u003e\n\u003cli\u003eXu X, Zhang X, Tian Q, Zhang G, Liu Y, Cui G, et al. Three-dimensional texture features from intensity and high-order derivative maps for the discrimination between bladder tumors and wall tissues via MRI. Int J Comput Assist Radiol Surg. 2017;12(4):645-56.\u003c/li\u003e\n\u003cli\u003eBrieu N, Gavriel CG, Nearchou IP, Harrison DJ, Schmidt G, Caie PD. Automated tumour budding quantification by machine learning augments TNM staging in muscle-invasive bladder cancer prognosis. Sci Rep. 2019;9(1):5174.\u003c/li\u003e\n\u003cli\u003eGalsky MD, Stensland K, Sfakianos JP, Mehrazin R, Diefenbach M, Mohamed N, et al. Comparative Effectiveness of Treatment Strategies for Bladder Cancer With Clinical Evidence of Regional Lymph Node Involvement. J Clin Oncol. 2016;34(22):2627-35.\u003c/li\u003e\n\u003cli\u003eJi J, Zhang T, Zhu L, Yao Y, Mei J, Sun L, et al. Using machine learning to develop preoperative model for lymph node metastasis in patients with bladder urothelial carcinoma. BMC Cancer. 2024;24(1):725.\u003c/li\u003e\n\u003cli\u003eMathieu R, Lucca I, Roupr\u0026ecirc;t M, Briganti A, Shariat SF. The prognostic role of lymphovascular invasion in urothelial carcinoma of the bladder. Nat Rev Urol. 2016;13(8):471-9.\u003c/li\u003e\n\u003cli\u003eTian YF, Zhou H, Yu G, Wang J, Li H, Xia D, et al. Prognostic significance of lymphovascular invasion in bladder cancer after surgical resection: A meta-analysis. J Huazhong Univ Sci Technolog Med Sci. 2015;35(5):646-55.\u003c/li\u003e\n\u003cli\u003eGarg T, Chen LY, Kim PH, Zhao PT, Herr HW, Donat SM. Preoperative serum albumin is associated with mortality and complications after radical cystectomy. BJU Int. 2014;113(6):918-23.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1.\u003c/strong\u003e \u003cstrong\u003eDemographic and Clinical Data\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"571\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eNo. Patients. n\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e241\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eAge (y); median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e65 (60-70)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 311px;\"\u003e\n \u003cp\u003eMale; n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e205 (85.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 311px;\"\u003e\n \u003cp\u003eFemale. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e36 (14.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eHydronephrosis. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e82 (34)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003ee-GFR before Cystectomy (ml/min/1.73 m\u003csup\u003e2\u003c/sup\u003e), median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e78 (52.5-91)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eACCI. median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e5 (4-6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 571px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClinical Stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003ecT1. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e17 (7.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003ecT2. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e118 (49)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003ecT3. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e74 (30.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003ecT4. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e32 (13.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eNeoadjuvant chemotherapy. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e41 (17.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 571px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCystectomy characteristics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" rowspan=\"6\" valign=\"top\" style=\"width: 155px;\"\u003e\n \u003cp\u003epT stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pT0. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e14 (5.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pTa. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e6 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pT1. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e24 (10)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pT2. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e58 (24.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pT3. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e75 (31.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pT4. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e62 (25.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" rowspan=\"4\" valign=\"top\" style=\"width: 155px;\"\u003e\n \u003cp\u003epN stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pN0. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e151 (62.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pN1. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e31 (12.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;pN2. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e58 (24.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 212px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;PN3. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e1 (0.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eVariant histology. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e55 (22.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eLymphovascular invasion. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e105 (43.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003ePositive surgical margin. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e15 (6.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eNecrosis. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e68 (28.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eAdjuvant chemotherapy. n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e42 (17.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eTur-Cystectomy interval time (mo), median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e2.1 (1.3-4.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eIntraoperative blood loss (ml), median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e1100 (100-6000)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eOperative time (min), median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e420 (300-690)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eLength of stay (d), median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e12 (7-56)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003ePeroperative Transfusion, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e216 (89.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eRS Pathology Cis, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e66 (27.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eAdenocarcinoma of the prostate. n (%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e60 (24.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 366px;\"\u003e\n \u003cp\u003eDeath, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e145 (60.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 571px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePreoperative blood parameters\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eHemoglobin, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e12 (10.8-13.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eCreatinine, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e1.2 (0.8-1.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eAlbumin, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e3.9 (3.4-4.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eTotal Protein, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e7 (6.5-7.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eNeutrophils, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e5.3 (3.9-7.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eLymphocytes, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e1.8 (1.3-2.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eMonocytes, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e0.61 (0.47-0.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003ePlatelet, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e235 (191.5-307.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 571px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eInflammatory parameters\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eNLR, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e2.8 (2.1-4.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eMLR, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e0.34 (0.25-0.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003ePLR, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e135.2(101.7-185)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eSII, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e715 (457-1108.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 313px;\"\u003e\n \u003cp\u003eSIRI, median (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 205px;\"\u003e\n \u003cp\u003e1.7 (1.1-3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 571px;\"\u003e\n \u003cp\u003eAbbreviations: e-GFR. estimated glomerular filtration rate; ACCI. age adjusted charlson comorbidity index; IQR. inter quartile rangeLVI, lymphovascular invasion; SIRI, systemic inflammatory response index; NLR, Neutrophil to lymphocyte ratio; MLR, Monocyte to lymphocyte ratio; PLR, Platelet to lymphocyte ratio; SII, Systemic inflammatory index\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2. Performance analysis of DS1 (30 original features) and DS-2 (original features plus all inflammatory markers) on experimental results of various models by evaluation metrics\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"621\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 280px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDS-1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"5\" valign=\"top\" style=\"width: 276px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDS-2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1-Score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrecision\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRecall\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eROC_AUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1-Score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrecision\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRecall\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eROC_AUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eCART\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,728565\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,751077\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,719047\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,6846\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,687394\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,731773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,766374\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,714761\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,693\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,701518\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eCatBoost\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,771186\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,744684\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,807142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,713166\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,771253\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,770551\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,76014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,784761\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,7175\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,747312\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eGBM\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,780757\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,754495\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,813333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,726166\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,769201\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,782028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,734484\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,840952\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,7176\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,757751\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eLightGBM\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,755676\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,768184\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,750476\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,7095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,71228\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,758028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,763158\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,758095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,7096\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,719285\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eRandomForests\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,784117\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,757227\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,820476\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,7298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,75773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,782969\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,725811\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,854285\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,7178\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,766013\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eXGBoost\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,785037\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,76984\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,807142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,7341\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,748216\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,774986\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,743152\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,813809\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,7178\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,737179\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eLogistic Regression\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,7464969\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,742808\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,758095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,6963\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,745444\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,762872\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,751794\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,778571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,713\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,74238\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eSVM\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,738954\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,750263\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,736666\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,6878\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,745925\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,746838\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,757916\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,743333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,7045\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,737074\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eMLP\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,773728\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0,762207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 52px;\"\u003e\n \u003cp\u003e0,792857\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,7255\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0,77301\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,762778\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 50px;\"\u003e\n \u003cp\u003e0,767999\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0,764761\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,7171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0,750465\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable.3: Changes with stepwise addition of inflammatory markers (SIRI, LMR, NLR, PLR, SII) to the original dataset on the evaluation of F1 Scores in a Random Forest model\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eusing DS-3.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"539\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCombinations\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1-Score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrecision\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRecall\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eROC_AUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003e\u003cem\u003eDS1+SIRI\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.791041\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.750858\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.840476\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.7341\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.786629\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003e\u003cem\u003eDS1+SIRI+LMR\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.788862\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.747563\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.841428\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.7301\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.775703\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003e\u003cem\u003eDS1+SIRI+NLR\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.793419\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.760625\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.835238\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.7385\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.779417\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eDS1+SIRI+NLR+PLR\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.801126\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.737930\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.883333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.7344\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.781582\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003e\u003cem\u003eDS1+SIRI+NLR+PLR+SII\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.789543\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.745874\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.841428\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e0.7301\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.754052\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"world-journal-of-urology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"wjur","sideBox":"Learn more about [World Journal of Urology](https://link.springer.com/journal/345)","snPcode":"345","submissionUrl":"https://submission.nature.com/new-submission/345/3","title":"World Journal of Urology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Machine Learning, Inflammatory markers, Radical Cystectomy, Shapley Additive Explanations (SHAP), Artificial intelligence","lastPublishedDoi":"10.21203/rs.3.rs-6717746/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6717746/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eWe aimed to create a Machine learning (ML) model using patient demographic, clinical and pathological data for prediction of overall survival in patients treated with radical cystectomy (RC). Secondly, we evaluated whether inflammatory markers contributed to this model.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe conducted a retrospective analysis of the institutional cystectomy database and identified consecutive RC patients. Dataset-1 (DS-1) was analyzed in ML models using 30 original features (including the target feature) encompassing preoperative, intraoperative, and postoperative data of the patients. All derived inflammatory markers were cumulatively added to DS-1 to create DS-2, and to test the specific contribution of inflammatory markers, they were systematically integrated in an ordinary order based on their predictive ability (DS-3). Markers without predictive contribution were excluded from the DS-3 model. In addition, the Shapley Additive Explanations (SHAP) method was used to examine the importance of each clinical feature and inflammatory marker.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eThe median age of the 241 patients was 65 years. The mortality rate was 60.2% (145/241). Two and 5-year overall survival (OS) rates were 54.7% and 37.2%, respectively. According to DS-1, F1 scores were between 0.72\u0026ndash;0.78. RandomForests and XGBoost models achieved the highest score of 0.78. DS-2 including all inflammatory markers but no significant improvement was found in F1 scores (0.73\u0026ndash;0.78). In DS-3, firstly adding the systemic inflammatory response index (SIRI) to the original features and then neutrophil/lymphocyte ratio (NLR) and platelet/lymphocyte ratio (PLR), achieved the highest F1 score (0.80) in the Random Forest model. SHAP analyses showed that Tumor (T) stage, preoperative albumin and presence of lympho-vascular invasion (LVI) contributed most to model predictivity.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eML models derived using demographic/clinical features resulted in a maximum F1 score of 0.78. However, adding the most predictive inflammatory markers in the sequence SIRI, NLR and PLR to demographic data achieved the highest F1 score of 0.80. Furthermore, T stage and preoperative albumin were the strongest predictive factors in the ML models.\u003c/p\u003e","manuscriptTitle":"Evaluation of Inflammatory Markers in Survival Analysis of Patients Undergoing Radical Cystectomy Using Machine Learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-03 13:23:37","doi":"10.21203/rs.3.rs-6717746/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-07-18T18:31:27+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-08T21:30:24+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-08T09:37:36+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-07T21:12:06+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-04T09:24:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"78657637709974756882178205738572811126","date":"2025-07-01T06:12:10+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"247808237925648166005874529327163910701","date":"2025-06-30T19:54:02+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"108366672574107664935570427588179889424","date":"2025-06-30T19:42:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"298215762453620425962131496093066572360","date":"2025-06-11T19:15:19+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-05-29T17:11:26+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-05-28T17:15:09+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-05-28T15:43:24+00:00","index":"","fulltext":""},{"type":"submitted","content":"World Journal of Urology","date":"2025-05-21T14:54:37+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"world-journal-of-urology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"wjur","sideBox":"Learn more about [World Journal of Urology](https://link.springer.com/journal/345)","snPcode":"345","submissionUrl":"https://submission.nature.com/new-submission/345/3","title":"World Journal of Urology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"0376ef5c-082a-4c03-b1da-00fd847d8768","owner":[],"postedDate":"June 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-10-27T16:25:57+00:00","versionOfRecord":{"articleIdentity":"rs-6717746","link":"https://doi.org/10.1007/s00345-025-05978-7","journal":{"identity":"world-journal-of-urology","isVorOnly":false,"title":"World Journal of Urology"},"publishedOn":"2025-10-21 16:16:39","publishedOnDateReadable":"October 21st, 2025"},"versionCreatedAt":"2025-06-03 13:23:37","video":"","vorDoi":"10.1007/s00345-025-05978-7","vorDoiUrl":"https://doi.org/10.1007/s00345-025-05978-7","workflowStages":[]},"version":"v1","identity":"rs-6717746","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6717746","identity":"rs-6717746","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}