Machine learning models predict the progression of long-term renal insufficiency in patients with renal cancer after radical nephrectomy

Machine learning models predict the progression of long-term renal insufficiency in patients with renal cancer after radical nephrectomy | 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 Machine learning models predict the progression of long-term renal insufficiency in patients with renal cancer after radical nephrectomy Yongchao Yan, Qihang Sun, Haotian Du, Yize Guo, Bin Li, Xinning Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5036531/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 18 Dec, 2024 Read the published version in BMC Nephrology → Version 1 posted 4 You are reading this latest preprint version Abstract Background: Chronic Kidney Disease (CKD) is a common severe complication after radical nephrectomy in patients with renal cancer. The timely and accurate prediction of the long-term progression of renal function post-surgery is crucial for early intervention and ultimately improving patient survival rates. Objective: This study aimed to establish a machine learning model to predict the likelihood of long-term renal dysfunction progression after surgery by analyzing patients’ general information in depth. Methods: We retrospectively collected data of eligible patients from the Affiliated Hospital of Qingdao University. The primary outcome was upgrading of the Chronic Kidney Disease stage between pre- and 3-year post-surgery. We constructed seven different machine-learning models based on Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (Lightgbm), Gaussian Naive Bayes (GaussianNB), and K-Nearest Neighbors (KNN). The performance of all predictive models was evaluated using the area under the receiver operating characteristic curve (AUC), precision-recall curves, confusion matrices, and calibration curves. Results: Among 360 patients with renal cancer who underwent radical nephrectomy included in this study, 185 (51.3%) experienced an upgrade in Chronic Kidney Disease stage 3-year post-surgery. Eleven predictive variables were selected for further construction of the machine learning models. The logistic regression model provided the most accurate prediction, with the highest AUC (0.8154) and an accuracy of 0.787. Conclusion: The logistic regression model can more accurately predict long-term renal dysfunction progression after radical nephrectomy in patients with renal cancer. Radical nephrectomy Chronic kidney disease Machine Learning Kidney cancer Early diagnosis Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Radical nephrectomy (RN) is an important surgical method for treating diseases such as renal cell carcinoma. Studies have shown that patients with renal cancer who have undergone RN are more prone to renal function deterioration and chronic kidney disease (CKD) ( 1 – 2 ). CKD is a common disease worldwide with a steadily increasing prevalence ( 3 ). Over the long term after kidney disease, there is also an increase in morbidity and mortality from cardiovascular and cerebrovascular diseases on the basis of CKD ( 4 ). However, many surgeons focus more on postoperative tumor outcomes than functional outcomes, and few articles study the long-term renal function changes after RN. Therefore, we can only refer to articles related to kidney transplant donors. However, Due to the strict selection criteria for donors in kidney transplant patients, the changes in kidney function in patients with kidney tumors are not representative. As expressed in the AUA and EUA guidelines, the follow-up of renal function in patients after Radical nephrectomy should be strengthened. The literature indicates that patients with a single kidney should be followed-up more strictly, as they are at a high risk of progressing to CKD ( 5 – 6 ). Therefore, we urgently need a clinical tool to predict the occurrence of CKD as well as the possibility of renal function deterioration in renal cancer patients after RN, so as to more clearly and effectively intervene with the target population in a timely manner, thereby improving patient prognosis. In recent years, machine learning has proven to be an effective tool for diagnosis, prognosis prediction, and interpretation of medical images ( 7 ). Because machine-learning models do not rely on the assumption of a linear relationship between input variables and outcomes, more available information can promote more precise and accurate prediction models. However, there is still a scarcity of definitive evidence regarding machine learning methods for predicting long-term renal function deterioration or CKD in patients with renal cancer after RN. With the continuous optimization of machine learning algorithms, we should develop efficient prediction models based on these algorithms to predict the likelihood of CKD or renal function deterioration in patients with renal cancer after RN in advance and to timely diagnose and treat perioperatively and in the early postoperative period. This can significantly improve the patient outcomes and survival rates. Therefore, our study aimed to find the most precise and reliable predictive model by establishing different machine learning models. Methods Study population This was a single-center retrospective study. We retrospectively analyzed the electronic medical records of 360 patients from the scientific research big data platform of Qingdao University Affiliated Hospital, China, from 2012 to 2020. Since this study was a secondary analysis of a publicly available database, approval from the Institutional Review Board (IRB) was not required. After completing the "Data or Specimens Only Research" course (Collaborative Institutional Training Initiative, CITI Program), we obtained access to the database. Inclusion and Exclusion Criteria All included patients had renal cell carcinoma and underwent radical nephrectomy from to 2012–2020. The exclusion criteria were as follows: 1) age three years; 3) Preoperative CKD stage greater than 3, and 4) incomplete clinical data. According to the CKD staging criteria proposed by the Kidney Disease Outcome Quality Initiative (KDOQI) of the National Kidney Foundation in 1999, the outcome of this study was whether CKD staging was upgraded. Data Extraction In this study, we included the following factors based on a literature review ( 4 , 8 – 10 ) and clinical experience: 1) demographic information (age, sex, and BMI); 2) comorbidity information (hypertension, diabetes, acute kidney injury, and intraoperative hypotension); 3) laboratory indicators (preoperative and postoperative creatinine ratio and preoperative eGFR); and 4) medical history information (surgical method and tumor size). Flowchart of Inclusion and Exclusion Criteria for Patients is shown in Fig. 1 . In this study, we randomly divided the patients into training and validation sets at a 7:3 ratio. We used the KDIGO clinical practice guidelines to recommend the CKD-EPI formula for estimating eGFR. We defined postoperative AKI according to the 2012 Kidney Disease: Improving Global Outcomes guidelines as an increase in plasma creatinine > 26.5 µmol/L within 48 hours or plasma creatinine > 1.5 times baseline, which is known or presumed to occur within 48 hours after surgery. Statistical Analysis Continuous variables are expressed as medians and interquartile ranges and compared using the Mann-Whitney U test. Categorical variables are presented as numbers and proportions and compared using the chi-square test or Fisher's exact test. Univariate and multivariate regression analyses were used to statistically analyze the selected variables and their relationships with the outcomes. SPSS software (version 25.0; IBM Corp., Armonk, NY, USA) was used for data analysis. Machine Learning Implementation We utilized seven machine learning algorithms to construct models for predicting the upgrading of the CKD stage: Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Gaussian Naive Bayes (GaussianNB), and K-Nearest Neighbors (KNN). Logistic Regression is a statistical model that models the probability of a dependent binary variable as a function of other variables ( 11 ). Support Vector Machine is a supervised machine learning model based on the Vapnik–Chervonenkis (VC) dimension and is capable of making robust predictions ( 12 ). It excels in performing both linear and nonlinear classifications. Random Forest is an ensemble learning method that builds multiple decision trees from training data and is classified by a majority voting mechanism ( 13 ). Extreme Gradient Boosting (XGBoost) is an ensemble machine-learning algorithm based on decision trees known for outstanding prediction performance and processing time ( 14 ). LightGBM is an efficient, distributed, and high-performance gradient boosting framework that boasts fast training speed, high accuracy, and scalability, and is suitable for large-scale and high-dimensional datasets for tasks such as regression, classification, and ranking ( 15 ). The Gaussian NB classifier is a Naive Bayes classifier based on Gaussian distribution. It assumes that the features are independent of each other and that each feature follows a Gaussian distribution, that is, a normal distribution ( 16 ). KNN is a classification algorithm based on the proximity to neighboring points. Its core idea is based on statistical theory, which classifies a point by measuring the weight of its neighbors and assigning it to the category with the greatest weight ( 17 ). The primary issue when conducting ensemble learning algorithms is avoiding overfitting. Fine-tuning the hyperparameters of the model is crucial to prevent such problems. In our study, the hyperparameters of all models were optimized to increase the area under the receiver operating characteristic (ROC) curve. The two requirements for hyperparameter tuning were: 1) the lowest training loss when trying different combinations of model hyperparameters; and 2) a logarithmic loss on the validation set smaller than -log0.5 (approximately 0.693), slightly higher than that on the training set. We provided the ROC curve and confusion matrix of each model to assess predictive performance. The ROC curve is a graphical tool that shows the diagnostic ability of a binary classifier at different discrimination thresholds by plotting the relationship between the true positive rate (TPR) and false positive rate (FPR). Furthermore, we present calibration curves and precision-recall curves to assess the predictive performance of all models. We compared the performance of each model in the training and validation sets in order to select the best predictive model. Results Baseline Characteristics From the research database of the Affiliated Hospital of Qingdao University, among 686 renal cancer patients who underwent radical nephrectomy between 2012 and 2020, 326 patients were excluded based on the following criteria: 18 patients were under 18 years of age, 214 patients had follow-up periods of less than three years; 36 patients had preoperative CKD staging > 3, and 58 patients had incomplete clinical data. Ultimately, 360 patients were included in the final analysis, of whom 51.3% experienced an upgrade in CKD stage three years post-surgery. Detailed information regarding the selected patients is provided in Table 1 . The baseline characteristics of the training and validation sets were similar. In the training and validation groups, the proportion of patients who experienced CKD stage progression three years post-surgery was 52.7% and 48.1%, respectively. The results of the univariate logistic regression and multivariate regression analyses are shown in Table 2 . We found that preoperative renal function (p < 0.001, OR = 1.036, EXP = 1.018–1.053), age (p < 0.001, OR = 1.050, EXP = 1.023–1.079), and the ratio of postoperative to preoperative creatinine (p = 0.0321, OR = 4.235, EXP = 1.131–15.859) were independent risk factors for CKD stage progression at three years post-surgery. Table 1. Demographic and baseline characteristics of the patients Variables Total(360) Train(252) Validation(108) Age, year Gender Male,n Female,n Hypertension, n Diabetes mellitus, n Surgical methods Open,n Laparoscopy,n Preoperative eGFR, mL/min/1.73 m2 BMI,Kg/m 2 Tumor size，cm the ratio of postoperative creatinine to preoperative creatinine Hypotensive shock,n AKI,n Upgrading CKD staging,36 months,n 58(51,65) 137(38.1%) 223(61.9%) 145(40.2%) 50(13.8%) 71(19.7%) 289(80.3%) 82.66(69.77,98.21) 24.8(23.0,27.2) 5(3.5,7) 1.27(1.09,1.48) 18(5.0%) 154(42.7%) 185(51.3%) 58(50,65) 94(37.3%) 158(62.6%) 108(42.8%) 39(15.4%) 51(20.2%) 201(79.8%) 81.29(68.51,97.24) 24.7(22.9,27.0) 5(3.5,7) 1.27(1.10,1.48) 11(4.3%) 109(43.2%) 133(52.7%) 57(52,65) 43(39.8%) 65(60.2%) 37(34.2%) 11(10.1%) 20(18.5%) 88(81.4%) 84.66(72.85,98.99) 25.2(23.0,27.5) 5(3.9,75) 1.24(1.07,1.47) 7(6.4%) 45(41.6%) 52(48.1%) AKI = acute kidney injury; eGFR = estimated glomerular filtration rate; BMI= Body Mass Index; AKI stages are divided according to KDIGO 2012 Table 2 Regression Analysis of influencing factors of long-term renal function Univariate analysis Multivariate analysis Variables OR 95% CI P-values OR 95% CI P-values Preoperative Renal function 1.034 1.021–1.047 < 0.001 1.036 1.018–1.053 < 0.001 Hypertension 0.740 0.485–1.130 0.163 0.690 0.410–1.163 0.164 Diabetes 1.171 0.644–2.128 0.606 1.405 0.689–2.865 0.350 Tumor size 0.955 0.892–1.022 0.185 0.975 0.891–1.067 0.582 Gender 1.561 1.017–2.394 0.042 1.371 0.834–2.254 0.214 BMI 0.988 0.949–1.028 0.549 0.975 0.931–1.020 0.273 Age 1.034 1.013–1.055 0.001 1.050 1.023–1.079 < 0.001 the ratio of postoperative creatinine to preoperative creatinine 10.677 5.011–22.748 < 0.001 4.235 1.131–15.859 0.032 Surgical methods 0.729 0.433–1.229 0.236 1.137 0.573–2.257 0.714 Hypotensive shock 0.512 0.188–1.395 0.190 0.560 0.174–1.801 0.331 AKI 0.398 0.259–0.612 < 0.001 1.094 0.513–2.332 0.816 Table.3 Testing the predictive performance of 7 models. Models Accuracy Precision Sensitivity (Recall) Specificity AUC LR 0.787 0.763 0.808 0.768 0.815 SVM 0.676 0.635 0.769 0.589 0.787 Lightgbm 0.750 0.736 0.750 0.750 0.810 RF 0.741 0.740 0.712 0.768 0.786 Xgboost 0.676 0.635 0.769 0.589 0.788 GaussianNB 0.639 0.686 0.462 0.804 0.744 KNN 0.667 0.629 0.75 0.589 0.729 Accuracy: The proportion of correctly classified samples to the total number of samples; Precision: also known as precision, it indicates the proportion of actual positive samples among the samples predicted as positive; Sensitivity (Recall): also known as recall rate, it indicates the proportion of actual positive samples in all positive samples among the predicted results; Specificity: specificity refers to the proportion of truly negative individuals correctly identified as negative through a diagnostic method in a non-diseased population. Evaluation of Model Performance We developed seven machine learning models to predict the occurrence of CKD stage progression three years after radical nephrectomy in patients with renal cancer. Table 3 summarizes the predictive performance of all developed machine learning models. Figure 2 displays the confusion matrices for each prediction model, and Fig. 3 shows each model's ROC curve and precision-recall curves. Among the seven models, the Logistic Regression model had the highest area under the ROC curve (AUC) of 0.8154. The calibration curves (Fig. 4 ) indicated that the probability of CKD progression predicted by the logistic regression model was in good agreement with the actual results. The importance rankings of the different predictive factors in this model are shown in Fig. 5 . Discussion Radical nephrectomy is a common surgical method for treating renal cell carcinoma in urology departments. As surgeons, the focus is often on tumor prognosis post-surgery, overlooking renal function prognosis. Chronic Kidney Disease (CKD) affects 8–16% of the global population, yet patients and clinicians often fail to fully recognize this condition ( 18 ). This results in a lack of early diagnosis and intervention for CKD, leading to increased morbidity and mortality rates ( 4 ). Therefore, we conducted this study to predict long-term progression of renal function in patients with renal cancer after radical nephrectomy. In current research, machine-learning models have been effectively applied in nephrology. The predictive performance of different machine-learning models varies across different diseases. Feng et al. developed an XGB machine learning model to predict lymph node metastasis in patients with renal cancer ( 19 ). Chen et al. used Support Vector Machine (SVM), Logistic Regression (LR), and Random Forest (RF) algorithms to build models for assessing renal function and fibrosis in diabetic nephropathy ( 20 ). Fernando Elihimas Júnior et al. applied a Logistic Regression model to predict poor renal function in elderly patients one year after kidney transplantation ( 21 ). In this study, we constructed and validated seven machine-learning models to predict long-term renal function progression in patients with renal carcinoma after radical nephrectomy. Among these machine-learning models, the logistic regression model demonstrated the best discriminative power and calibration ability. Logistic regression is a classic machine learning algorithm used to solve binary classification problems. In machine learning, logistic regression is a type of supervised learning algorithm that is primarily used to predict the relationship between input variables and a binary output variable. It is used to map a linear combination of input features through a Logistic Function (logistic function) to a probability value, which in binary classification problems is usually expressed as the probability of the category being 1. During the training process, logistic regression learns the best model parameters by maximizing the likelihood function or minimizing the loss function ( 22 ). This study is the first to use machine learning models to predict long-term renal function progression in patients with renal carcinoma after radical nephrectomy. This is important for early diagnosis and management of CKD. As kidney damage is irreversible, it is important to intervene and manage high-risk populations early in clinical practice. For example, the early management of diabetes is crucial. Blood sugar control can delay the progression of CKD, and most guidelines suggest a target glycated hemoglobin A1c level of ~ 7.0% ( 23 , 24 ). Medication dosage and selection should also receive important attention for populations at a high risk of developing CKD. All patients at high risk for CKD should be advised to avoid nephrotoxins. While a complete list is beyond the scope of this article, there are several points worth mentioning. Chronic kidney disease Patients with CKD are advised not to routinely use nonsteroidal anti-inflammatory drugs, particularly those receiving ACE-I or ARB treatment ( 23 ). Common medications that require dosage reduction include antibiotics, direct oral anticoagulants, gabapentin and pregabalin, oral hypoglycemic agents, insulin, chemotherapy drugs, and opioids ( 23 ). Therefore, early dietary management is important. The KDIGO guidelines recommend that in adults at risk for CKD, reducing protein intake to below 1.3 g/kg per day may benefit from progressively restricted dietary protein, which must be balanced against the concerns of malnutrition and/or protein catabolic syndrome ( 25 ). Lowering the dietary acid load (e.g., eating more fruits and vegetables and less meat, eggs, and cheese) can also help prevent kidney damage ( 26 ). Our study has yielded an interesting conclusion. Preoperative renal function and the ratio of postoperative to preoperative creatinine levels are independent risk factors for the occurrence of CKD or deterioration of renal function in postoperative patients. The higher the preoperative renal function level, the higher is the risk of long-term deterioration of renal function. A larger creatinine ratio indicates more renal function loss after surgery than before surgery, thus increasing the risk of long-term renal function deterioration. Moreover, these two factors were the most important among all the elements in this prediction model, ranking as the top two. Olcucuoglu et al. indicated that patients who undergo radical nephrectomy might experience a more rapid deterioration of renal function than those who gradually lose kidney function ( 1 ). This may be due to the loss of nephron units in one kidney, leading to hyperfiltration and proliferation of the remaining nephron units as a compensatory hypertrophy of the remaining kidney ( 27 – 29 ). Initially, maintaining the GFR at a certain rate may seem beneficial, but it leads to specific structural changes, such as glomerulosclerosis and tubular atrophy. Clinically, these changes manifest as hypertension (HT), proteinuria, and GFR decline. HT and glomerulosclerosis eventually cause further nephron loss, triggering CKD development ( 30 ). This interesting finding may be counterintuitive, as it is often thought that the better the kidney function, the better is the reserve function. Therefore, this reminds us that in clinical practice, we should pay more attention to patients with initially very healthy kidney tumors, indicating that their physiological function may decline more severely after surgery. However, our study had some limitations. First, these machine-learning models are based on single-center data for training and development, requiring further external validation to interpret the universality of these models. Second, this was a retrospective and observational study; therefore, bias was inevitable. Third, future prospective studies with larger sample sizes are necessary to further validate the potential of machine learning models to predict clinical outcomes. Conclusion This study found that the better the preoperative renal function, the greater the ratio of postoperative to preoperative creatinine; the older the age, the more likely it is that renal function will deteriorate after radical nephrectomy. The logistic regression model based on patient clinical data showed the best predictive performance. It can assist clinicians in the early assessment of the risk of long-term renal function deterioration or development into CKD post-surgery, allowing for early intervention and improving patient prognosis and quality of life. Declarations Funding Statement: N/A Conflict of Interest declaration: The authors declare that they have NO affiliations with or involvement in any organization or entity with any financial interest in the subject matter or materials discussed in this manuscript. The ethical approval number is QYFY WZLL 28808. Author Contributions: Yongchao Yan to the analysis of the results and to the writing of the manuscript. Qihang Sun and Haotian Du contributed to the design and implementation of the research, Yize Guo and Bin Li were involved in collecting data. Xinning Wang conceived the original and supervised the project. Ethics approval and consent to participate: As this publication is a report that contains no identifiable content to the patient, this publication was exempt from ethical approval by the Human Research Protection Program (HRPP) and its Institutional Review Board (IRB) at the Ethics Committee of the affiliated hospital of Qingdao University. Consent for publication: N/A Availability of data and materials: The datasets used and/or analysed during the current study available from the corresponding author on reasonable request. Competing interest: The author declare that there no competing interests. Funding: This research was supported by the Shandong Province medical health science and technology project (NO.202304051689). Animal Studies: N/A Registry and the Registration No. of the study/trial: N/A Acknowledgments: N/A References Olcucuoglu E, Tonyali S, Tastemur S, Kasap Y, Sirin ME, Gazel E, et al. Comparison of long-term kidney functions after radical nephrectomy and simple nephrectomy. PeerJ. 2019;7:e6701. Guidi E, Cozzi M, Milani S, Spada E. Nephrectomy modifies renal angiotensin II effects in kidney donors. Am J Hypertens. 2008;21(5):592-8. Qiu Y, Zhao Q, Wang N, Yu Y, Wang R, Zhang Y, et al. Association of hypertriglyceridemic waist phenotype with renal function impairment: a cross-sectional study in a population of Chinese adults. Nutr Metab (Lond). 2020;17:63. Zabell J, Demirjian S, Lane BR, Derweesh IH, Isharwal S, Suk-Ouichai C, et al. Predictors of Long-Term Survival after Renal Cancer Surgery. J Urol. 2018;199(2):384-92. Charles C, Ferris AH. Chronic Kidney Disease. Prim Care. 2020;47(4):585-95. Schreuder MF. Life with one kidney. Pediatr Nephrol. 2018;33(4):595-604. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, et al. Artificial intelligence in healthcare: past, present and future. Stroke and Vascular Neurology. 2017;2(4):230-43. Zabor EC, Furberg H, Lee B, Campbell S, Lane BR, Thompson RH, et al. Long-Term Renal Function Recovery following Radical Nephrectomy for Kidney Cancer: Results from a Multicenter Confirmatory Study. J Urol. 2018;199(4):921-6. Roussel E, Laenen A, Bhindi B, De Dobbeleer A, Stichele AV, Verbeke L, et al. Predicting short- and long-term renal function following partial and radical nephrectomy. Urol Oncol. 2023;41(2):110 e1- e6. Laguna MP. Re: Renal function after nephron-sparing surgery versus radical nephrectomy: results from EORTC randomized trial 30904. J Urol. 2014;192(2):369-70. Stoltzfus JC. Logistic regression: a brief primer. Acad Emerg Med. 2011;18(10):1099-1104. doi:10.1111/j.1553-2712.2011.01185.x Valkenborg D, Rousseau AJ, Geubbelmans M, Burzykowski T. Support vector machines. Am J Orthod Dentofacial Orthop. 2023;164(5):754-7. Rigatti SJ. Random Forest. J Insur Med. 2017;47(1):31-39. doi:10.17849/insm-47-01-31-39.1. Koh J. Gradient boosting with extreme-value theory for wildfire prediction. Extremes (Boston). 2023;26(2):273-99. Liao H, Zhang X, Zhao C, Chen Y, Zeng X, Li H. LightGBM: an efficient and accurate method for predicting pregnancy diseases. J Obstet Gynaecol. 2022;42(4):620-9. Ontivero-Ortega M, Lage-Castellanos A, Valente G, Goebel R, Valdes-Sosa M. Fast Gaussian Naive Bayes for searchlight classification analysis. Neuroimage. 2017;163:471-9. Arian R, Hariri A, Mehridehnavi A, Fassihi A, Ghasemi F. Protein kinase inhibitors' classification using K-Nearest neighbor algorithm. Comput Biol Chem. 2020;86:107269. Chen TK, Knicely DH, Grams ME. Chronic Kidney Disease Diagnosis and Management: A Review. JAMA. 2019;322(13):1294-304. Feng X, Hong T, Liu W, Xu C, Li W, Yang B, et al. Development and validation of a machine learning model to predict the risk of lymph node metastasis in renal carcinoma. Front Endocrinol (Lausanne). 2022;13:1054358. Chen W, Zhang L, Cai G, Zhang B, Lian Z, Li J, et al. Machine learning-based multimodal MRI texture analysis for assessing renal function and fibrosis in diabetic nephropathy: a retrospective study. Front Endocrinol (Lausanne). 2023;14:1050078. Elihimas Junior UF, Couto JP, Pereira W, Barros de Oliveira Sa MP, Tenorio de Franca EE, Aguiar FC, et al. Logistic Regression Model in a Machine Learning Application to Predict Elderly Kidney Transplant Recipients with Worse Renal Function One Year after Kidney Transplant: Elderly KTbot. J Aging Res. 2020;2020:7413616. Stoltzfus JC. Logistic regression: a brief primer. Acad Emerg Med. 2011;18(10):1099-104. Inker LA, Astor BC, Fox CH, Isakova T, Lash JP, Peralta CA, et al. KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD. Am J Kidney Dis. 2014;63(5):713-35. Guideline development g. Clinical Practice Guideline on management of patients with diabetes and chronic kidney disease stage 3b or higher (eGFR <45 mL/min). Nephrol Dial Transplant. 2015;30 Suppl 2:ii1-142. Aparicio M, Fouque D, Chauveau P. Effect of a very low-protein diet on long-term outcomes. Am J Kidney Dis. 2009;54(1):183. Goraya N, Simoni J, Jo C, Wesson DE. Dietary acid reduction with fruits and vegetables or bicarbonate attenuates kidney injury in patients with a moderately reduced glomerular filtration rate due to hypertensive nephropathy. Kidney Int. 2012;81(1):86-93. Kasiske BL, Ma JZ, Louis TA, Swan SK. Long-term effects of reduced renal mass in humans. Kidney Int. 1995;48(3):814-819. doi:10.1038/ki.1995.355. Brenner BM, Lawler EV, Mackenzie HS. The hyperfiltration theory: a paradigm shift in nephrology. Kidney Int. 1996;49(6):1774-1777. doi:10.1038/ki.1996.265. Helal I, Fick-Brosnahan GM, Reed-Gitomer B, Schrier RW. Glomerular hyperfiltration: definitions, mechanisms and clinical implications. Nat Rev Nephrol. 2012;8(5):293-300. Schreuder MF, Langemeijer ME, Bokenkamp A, Delemarre-Van de Waal HA, Van Wijk JA. Hypertension and microalbuminuria in children with congenital solitary kidneys. J Paediatr Child Health. 2008;44(6):363-8. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 18 Dec, 2024 Read the published version in BMC Nephrology → Version 1 posted Editorial decision: Revision requested 13 Sep, 2024 Editor assigned by journal 13 Sep, 2024 Submission checks completed at journal 13 Sep, 2024 First submitted to journal 05 Sep, 2024 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-5036531","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":353653660,"identity":"36ac6536-842a-466e-95df-3a9f3e433197","order_by":0,"name":"Yongchao Yan","email":"","orcid":"","institution":"The Affiliated Hospital of Qingdao University","correspondingAuthor":false,"prefix":"","firstName":"Yongchao","middleName":"","lastName":"Yan","suffix":""},{"id":353653661,"identity":"c587935e-674d-4df5-98b8-e0c14b87f954","order_by":1,"name":"Qihang Sun","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Qihang","middleName":"","lastName":"Sun","suffix":""},{"id":353653662,"identity":"2b9f5011-5587-49ff-8f41-f48800cc0348","order_by":2,"name":"Haotian Du","email":"","orcid":"","institution":"The Affiliated Hospital of Qingdao University","correspondingAuthor":false,"prefix":"","firstName":"Haotian","middleName":"","lastName":"Du","suffix":""},{"id":353653663,"identity":"13239cbb-5e81-49f0-8b1b-dd4045482c88","order_by":3,"name":"Yize Guo","email":"","orcid":"","institution":"The Affiliated Hospital of Qingdao University","correspondingAuthor":false,"prefix":"","firstName":"Yize","middleName":"","lastName":"Guo","suffix":""},{"id":353653664,"identity":"8fa913f5-4eeb-4854-89ae-cadba4a5af54","order_by":4,"name":"Bin Li","email":"","orcid":"","institution":"The Affiliated Hospital of Qingdao University","correspondingAuthor":false,"prefix":"","firstName":"Bin","middleName":"","lastName":"Li","suffix":""},{"id":353653665,"identity":"d554a3ff-2484-418f-9f8f-397a7e825b5f","order_by":5,"name":"Xinning Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAu0lEQVRIiWNgGAWjYJCCAx8qbOTY2JsPEK2D8eCMM2nGfDzHEojWwnyYt+1w4jyJHAXi1BvcyD0A1MKc3saQw8Dwo2IbEVrOnEs4OOccW24bw9kDjD1nbhPWYna8x+DAmzKe3DbGvgRmxjZitBzmMTjAwyaRzsbMY0CkFqAtB3naDBLY2IjVYn/mjAEwkBMM23jYEg4S5RfJGTnGHz5U/JeXn//44IMfFURoQQEHSFQ/CkbBKBgFowAXAADBI0AQabtPXgAAAABJRU5ErkJggg==","orcid":"","institution":"The Affiliated Hospital of Qingdao University","correspondingAuthor":true,"prefix":"","firstName":"Xinning","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2024-09-05 08:23:54","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5036531/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5036531/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12882-024-03907-1","type":"published","date":"2024-12-18T15:57:27+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":70506719,"identity":"751d55bd-daec-4278-85fe-0fde27bb14d9","added_by":"auto","created_at":"2024-12-03 23:41:18","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":84182,"visible":true,"origin":"","legend":"\u003cp\u003ePatient Screening Process Flowchart\u003c/p\u003e","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5036531/v1/38a44ce5b000cf7d233415b5.jpg"},{"id":70506722,"identity":"b542a333-d317-41c5-b9d1-33a82b4ff56c","added_by":"auto","created_at":"2024-12-03 23:41:18","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":124872,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion matrix of the machine learning model. a: Logistic Regression (LR); b: Support Vector Machine (SVM); c: Random Forest (RF); d: Extreme Gradient Boosting (XGBoost); e: Light Gradient Boosting Machine (Lightgbm); f: Gaussian Naive Bayes (GaussianNB); g: K-Nearest Neighbors (KNN).\u003c/p\u003e","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5036531/v1/ead917af1fd953c2838b7a25.jpg"},{"id":70506093,"identity":"2fea5499-6ac6-4323-bfca-33559b53c72e","added_by":"auto","created_at":"2024-12-03 23:33:18","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":103648,"visible":true,"origin":"","legend":"\u003cp\u003eEach model's ROC curve.\u003c/p\u003e","description":"","filename":"Figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5036531/v1/169719fd82842ac299cd9c4a.jpg"},{"id":70506099,"identity":"214ef92a-4a98-4107-b743-cc6b3a73f53f","added_by":"auto","created_at":"2024-12-03 23:33:18","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":147450,"visible":true,"origin":"","legend":"\u003cp\u003eEach model's precision-recall curves.\u003c/p\u003e","description":"","filename":"Figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5036531/v1/913f9bf3326d27f3d1ad131c.jpg"},{"id":70506098,"identity":"6f7134f4-921b-4055-9770-4b113ae9dd72","added_by":"auto","created_at":"2024-12-03 23:33:18","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":184498,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curves of seven machine learning models based on the validation set.\u003c/p\u003e","description":"","filename":"Figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5036531/v1/d6c8a383644cefaf77388424.jpg"},{"id":70507021,"identity":"d6daa373-c2b4-459a-8349-9afe1f9944ef","added_by":"auto","created_at":"2024-12-03 23:49:18","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":62362,"visible":true,"origin":"","legend":"\u003cp\u003eThe importance rankings of the different predictive factors in this model\u003c/p\u003e","description":"","filename":"Figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5036531/v1/131e629ce6391e81dcb81702.jpg"},{"id":72201704,"identity":"0352fbe6-618f-4960-8f4c-cdd2e0a88963","added_by":"auto","created_at":"2024-12-23 16:10:01","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1192881,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5036531/v1/4a1e81dc-6b78-4a4e-ad36-93fae6b7fd2a.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine learning models predict the progression of long-term renal insufficiency in patients with renal cancer after radical nephrectomy","fulltext":[{"header":"Introduction","content":"\u003cp\u003eRadical nephrectomy (RN) is an important surgical method for treating diseases such as renal cell carcinoma. Studies have shown that patients with renal cancer who have undergone RN are more prone to renal function deterioration and chronic kidney disease (CKD) (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e–\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). CKD is a common disease worldwide with a steadily increasing prevalence (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Over the long term after kidney disease, there is also an increase in morbidity and mortality from cardiovascular and cerebrovascular diseases on the basis of CKD (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). However, many surgeons focus more on postoperative tumor outcomes than functional outcomes, and few articles study the long-term renal function changes after RN. Therefore, we can only refer to articles related to kidney transplant donors. However, Due to the strict selection criteria for donors in kidney transplant patients, the changes in kidney function in patients with kidney tumors are not representative.\u003c/p\u003e \u003cp\u003e As expressed in the AUA and EUA guidelines, the follow-up of renal function in patients after Radical nephrectomy should be strengthened. The literature indicates that patients with a single kidney should be followed-up more strictly, as they are at a high risk of progressing to CKD (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e–\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). Therefore, we urgently need a clinical tool to predict the occurrence of CKD as well as the possibility of renal function deterioration in renal cancer patients after RN, so as to more clearly and effectively intervene with the target population in a timely manner, thereby improving patient prognosis.\u003c/p\u003e \u003cp\u003eIn recent years, machine learning has proven to be an effective tool for diagnosis, prognosis prediction, and interpretation of medical images (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Because machine-learning models do not rely on the assumption of a linear relationship between input variables and outcomes, more available information can promote more precise and accurate prediction models. However, there is still a scarcity of definitive evidence regarding machine learning methods for predicting long-term renal function deterioration or CKD in patients with renal cancer after RN.\u003c/p\u003e \u003cp\u003eWith the continuous optimization of machine learning algorithms, we should develop efficient prediction models based on these algorithms to predict the likelihood of CKD or renal function deterioration in patients with renal cancer after RN in advance and to timely diagnose and treat perioperatively and in the early postoperative period. This can significantly improve the patient outcomes and survival rates. Therefore, our study aimed to find the most precise and reliable predictive model by establishing different machine learning models.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e "},{"header":"Methods","content":"\u003cp\u003eStudy population\u003c/p\u003e\u003cp\u003eThis was a single-center retrospective study. We retrospectively analyzed the electronic medical records of 360 patients from the scientific research big data platform of Qingdao University Affiliated Hospital, China, from 2012 to 2020. Since this study was a secondary analysis of a publicly available database, approval from the Institutional Review Board (IRB) was not required. After completing the \"Data or Specimens Only Research\" course (Collaborative Institutional Training Initiative, CITI Program), we obtained access to the database.\u003c/p\u003e\u003cp\u003eInclusion and Exclusion Criteria\u003c/p\u003e\u003cp\u003eAll included patients had renal cell carcinoma and underwent radical nephrectomy from to 2012–2020. The exclusion criteria were as follows: 1) age \u0026lt; 18 years, 2) follow-up time \u0026gt; three years; 3) Preoperative CKD stage greater than 3, and 4) incomplete clinical data. According to the CKD staging criteria proposed by the Kidney Disease Outcome Quality Initiative (KDOQI) of the National Kidney Foundation in 1999, the outcome of this study was whether CKD staging was upgraded.\u003c/p\u003e\u003cp\u003eData Extraction\u003c/p\u003e\u003cp\u003eIn this study, we included the following factors based on a literature review (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e–\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e) and clinical experience: 1) demographic information (age, sex, and BMI); 2) comorbidity information (hypertension, diabetes, acute kidney injury, and intraoperative hypotension); 3) laboratory indicators (preoperative and postoperative creatinine ratio and preoperative eGFR); and 4) medical history information (surgical method and tumor size). Flowchart of Inclusion and Exclusion Criteria for Patients is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. In this study, we randomly divided the patients into training and validation sets at a 7:3 ratio. We used the KDIGO clinical practice guidelines to recommend the CKD-EPI formula for estimating eGFR. We defined postoperative AKI according to the 2012 Kidney Disease: Improving Global Outcomes guidelines as an increase in plasma creatinine \u0026gt; 26.5 µmol/L within 48 hours or plasma creatinine \u0026gt; 1.5 times baseline, which is known or presumed to occur within 48 hours after surgery.\u003c/p\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eContinuous variables are expressed as medians and interquartile ranges and compared using the Mann-Whitney U test. Categorical variables are presented as numbers and proportions and compared using the chi-square test or Fisher's exact test. Univariate and multivariate regression analyses were used to statistically analyze the selected variables and their relationships with the outcomes. SPSS software (version 25.0; IBM Corp., Armonk, NY, USA) was used for data analysis.\u003c/p\u003e\u003cp\u003eMachine Learning Implementation\u003c/p\u003e\u003cp\u003eWe utilized seven machine learning algorithms to construct models for predicting the upgrading of the CKD stage: Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Gaussian Naive Bayes (GaussianNB), and K-Nearest Neighbors (KNN). Logistic Regression is a statistical model that models the probability of a dependent binary variable as a function of other variables (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Support Vector Machine is a supervised machine learning model based on the Vapnik–Chervonenkis (VC) dimension and is capable of making robust predictions (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). It excels in performing both linear and nonlinear classifications. Random Forest is an ensemble learning method that builds multiple decision trees from training data and is classified by a majority voting mechanism (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Extreme Gradient Boosting (XGBoost) is an ensemble machine-learning algorithm based on decision trees known for outstanding prediction performance and processing time (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). LightGBM is an efficient, distributed, and high-performance gradient boosting framework that boasts fast training speed, high accuracy, and scalability, and is suitable for large-scale and high-dimensional datasets for tasks such as regression, classification, and ranking (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). The Gaussian NB classifier is a Naive Bayes classifier based on Gaussian distribution. It assumes that the features are independent of each other and that each feature follows a Gaussian distribution, that is, a normal distribution (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). KNN is a classification algorithm based on the proximity to neighboring points. Its core idea is based on statistical theory, which classifies a point by measuring the weight of its neighbors and assigning it to the category with the greatest weight (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). The primary issue when conducting ensemble learning algorithms is avoiding overfitting. Fine-tuning the hyperparameters of the model is crucial to prevent such problems. In our study, the hyperparameters of all models were optimized to increase the area under the receiver operating characteristic (ROC) curve. The two requirements for hyperparameter tuning were: 1) the lowest training loss when trying different combinations of model hyperparameters; and 2) a logarithmic loss on the validation set smaller than -log0.5 (approximately 0.693), slightly higher than that on the training set. We provided the ROC curve and confusion matrix of each model to assess predictive performance. The ROC curve is a graphical tool that shows the diagnostic ability of a binary classifier at different discrimination thresholds by plotting the relationship between the true positive rate (TPR) and false positive rate (FPR). Furthermore, we present calibration curves and precision-recall curves to assess the predictive performance of all models. We compared the performance of each model in the training and validation sets in order to select the best predictive model.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eBaseline Characteristics\u003c/p\u003e\n\u003cp\u003eFrom the research database of the Affiliated Hospital of Qingdao University, among 686 renal cancer patients who underwent radical nephrectomy between 2012 and 2020, 326 patients were excluded based on the following criteria: 18 patients were under 18 years of age, 214 patients had follow-up periods of less than three years; 36 patients had preoperative CKD staging\u0026thinsp;\u0026gt;\u0026thinsp;3, and 58 patients had incomplete clinical data. Ultimately, 360 patients were included in the final analysis, of whom 51.3% experienced an upgrade in CKD stage three years post-surgery. Detailed information regarding the selected patients is provided in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. The baseline characteristics of the training and validation sets were similar. In the training and validation groups, the proportion of patients who experienced CKD stage progression three years post-surgery was 52.7% and 48.1%, respectively. The results of the univariate logistic regression and multivariate regression analyses are shown in Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e. We found that preoperative renal function (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, OR\u0026thinsp;=\u0026thinsp;1.036, EXP\u0026thinsp;=\u0026thinsp;1.018\u0026ndash;1.053), age (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, OR\u0026thinsp;=\u0026thinsp;1.050, EXP\u0026thinsp;=\u0026thinsp;1.023\u0026ndash;1.079), and the ratio of postoperative to preoperative creatinine (p\u0026thinsp;=\u0026thinsp;0.0321, OR\u0026thinsp;=\u0026thinsp;4.235, EXP\u0026thinsp;=\u0026thinsp;1.131\u0026ndash;15.859) were independent risk factors for CKD stage progression at three years post-surgery.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" align=\"left\" width=\"101%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eTable 1. Demographic and baseline characteristics of the patients\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003eTotal(360)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003eTrain(252)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003eValidation(108)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003eAge, year\u003c/p\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003cp\u003eMale,n\u003c/p\u003e\n \u003cp\u003eFemale,n\u003c/p\u003e\n \u003cp\u003eHypertension, n\u003c/p\u003e\n \u003cp\u003eDiabetes mellitus, n\u003c/p\u003e\n \u003cp\u003eSurgical methods\u003c/p\u003e\n \u003cp\u003eOpen,n\u003c/p\u003e\n \u003cp\u003eLaparoscopy,n\u003c/p\u003e\n \u003cp\u003ePreoperative eGFR, mL/min/1.73 m2 \u0026nbsp;\u003c/p\u003e\n \u003cp\u003eBMI,Kg/m\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003cp\u003eTumor size，cm\u003c/p\u003e\n \u003cp\u003ethe ratio of postoperative creatinine to preoperative creatinine\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eHypotensive shock,n\u003c/p\u003e\n \u003cp\u003eAKI,n\u003c/p\u003e\n \u003cp\u003eUpgrading CKD staging,36 months,n\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e58(51,65) \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e137(38.1%)\u003c/p\u003e\n \u003cp\u003e223(61.9%)\u003c/p\u003e\n \u003cp\u003e145(40.2%)\u003c/p\u003e\n \u003cp\u003e50(13.8%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e71(19.7%)\u003c/p\u003e\n \u003cp\u003e289(80.3%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e82.66(69.77,98.21)\u003c/p\u003e\n \u003cp\u003e24.8(23.0,27.2)\u003c/p\u003e\n \u003cp\u003e5(3.5,7)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e1.27(1.09,1.48)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e18(5.0%)\u003c/p\u003e\n \u003cp\u003e154(42.7%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e185(51.3%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; 58(50,65)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e94(37.3%)\u003c/p\u003e\n \u003cp\u003e158(62.6%)\u003c/p\u003e\n \u003cp\u003e108(42.8%)\u003c/p\u003e\n \u003cp\u003e39(15.4%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e51(20.2%)\u003c/p\u003e\n \u003cp\u003e201(79.8%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e81.29(68.51,97.24)\u003c/p\u003e\n \u003cp\u003e24.7(22.9,27.0)\u003c/p\u003e\n \u003cp\u003e5(3.5,7)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e1.27(1.10,1.48)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e11(4.3%)\u003c/p\u003e\n \u003cp\u003e109(43.2%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e133(52.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22px;\"\u003e\n \u003cp\u003e57(52,65)\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e43(39.8%)\u003c/p\u003e\n \u003cp\u003e65(60.2%)\u003c/p\u003e\n \u003cp\u003e37(34.2%)\u003c/p\u003e\n \u003cp\u003e11(10.1%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e20(18.5%)\u003c/p\u003e\n \u003cp\u003e88(81.4%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e84.66(72.85,98.99)\u003c/p\u003e\n \u003cp\u003e25.2(23.0,27.5)\u003c/p\u003e\n \u003cp\u003e5(3.9,75)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e1.24(1.07,1.47)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e7(6.4%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;45(41.6%)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; 52(48.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eAKI = acute kidney injury; eGFR = estimated glomerular filtration rate; BMI= Body Mass Index; AKI stages are divided according to KDIGO 2012\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cdiv\u003e\u003cbr\u003e\u003c/div\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eRegression Analysis of influencing factors of long-term renal function\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003eUnivariate analysis\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003eMultivariate analysis\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eP-values\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eP-values\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePreoperative Renal function\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.034\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.021\u0026ndash;1.047\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.036\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.018\u0026ndash;1.053\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.740\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.485\u0026ndash;1.130\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.163\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.410\u0026ndash;1.163\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.164\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDiabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.644\u0026ndash;2.128\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.606\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.405\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.689\u0026ndash;2.865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.350\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor size\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.955\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.892\u0026ndash;1.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.185\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.975\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.891\u0026ndash;1.067\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.582\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.561\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.017\u0026ndash;2.394\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.042\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.371\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.834\u0026ndash;2.254\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.214\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.988\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.949\u0026ndash;1.028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.549\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.975\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.931\u0026ndash;1.020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.273\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.034\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.013\u0026ndash;1.055\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.050\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.023\u0026ndash;1.079\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ethe ratio of postoperative creatinine to preoperative creatinine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10.677\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.011\u0026ndash;22.748\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.235\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.131\u0026ndash;15.859\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSurgical methods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.729\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.433\u0026ndash;1.229\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.236\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.137\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.573\u0026ndash;2.257\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.714\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHypotensive shock\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.512\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.188\u0026ndash;1.395\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.190\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.560\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.174\u0026ndash;1.801\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.331\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAKI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.398\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.259\u0026ndash;0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.094\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.513\u0026ndash;2.332\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.816\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Taba\" border=\"1\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colspan=\"6\"\u003e\n \u003cp\u003eTable.3 Testing the predictive performance of 7 models.\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eModels\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003cp\u003e(Recall)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.787\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.763\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.808\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.768\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.815\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.635\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.769\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.589\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.787\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLightgbm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.736\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.810\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.741\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.740\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.712\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.768\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.786\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eXgboost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.635\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.769\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.589\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.788\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGaussianNB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.639\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.686\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.462\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.744\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKNN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.667\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.629\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.589\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.729\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"6\"\u003e\n \u003cp\u003eAccuracy: The proportion of correctly classified samples to the total number of samples; Precision: also known as precision, it indicates the proportion of actual positive samples among the samples predicted as positive; Sensitivity (Recall): also known as recall rate, it indicates the proportion of actual positive samples in all positive samples among the predicted results; Specificity: specificity refers to the proportion of truly negative individuals correctly identified as negative through a diagnostic method in a non-diseased population.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eEvaluation of Model Performance\u003c/p\u003e\n\u003cp\u003eWe developed seven machine learning models to predict the occurrence of CKD stage progression three years after radical nephrectomy in patients with renal cancer. Table\u0026nbsp;3 summarizes the predictive performance of all developed machine learning models. Figure\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e displays the confusion matrices for each prediction model, and Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e shows each model\u0026apos;s ROC curve and precision-recall curves. Among the seven models, the Logistic Regression model had the highest area under the ROC curve (AUC) of 0.8154. The calibration curves (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e) indicated that the probability of CKD progression predicted by the logistic regression model was in good agreement with the actual results. The importance rankings of the different predictive factors in this model are shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eRadical nephrectomy is a common surgical method for treating renal cell carcinoma in urology departments. As surgeons, the focus is often on tumor prognosis post-surgery, overlooking renal function prognosis. Chronic Kidney Disease (CKD) affects 8\u0026ndash;16% of the global population, yet patients and clinicians often fail to fully recognize this condition (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). This results in a lack of early diagnosis and intervention for CKD, leading to increased morbidity and mortality rates (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). Therefore, we conducted this study to predict long-term progression of renal function in patients with renal cancer after radical nephrectomy.\u003c/p\u003e \u003cp\u003eIn current research, machine-learning models have been effectively applied in nephrology. The predictive performance of different machine-learning models varies across different diseases. Feng et al. developed an XGB machine learning model to predict lymph node metastasis in patients with renal cancer (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). Chen et al. used Support Vector Machine (SVM), Logistic Regression (LR), and Random Forest (RF) algorithms to build models for assessing renal function and fibrosis in diabetic nephropathy (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). Fernando Elihimas J\u0026uacute;nior et al. applied a Logistic Regression model to predict poor renal function in elderly patients one year after kidney transplantation (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn this study, we constructed and validated seven machine-learning models to predict long-term renal function progression in patients with renal carcinoma after radical nephrectomy. Among these machine-learning models, the logistic regression model demonstrated the best discriminative power and calibration ability. Logistic regression is a classic machine learning algorithm used to solve binary classification problems. In machine learning, logistic regression is a type of supervised learning algorithm that is primarily used to predict the relationship between input variables and a binary output variable. It is used to map a linear combination of input features through a Logistic Function (logistic function) to a probability value, which in binary classification problems is usually expressed as the probability of the category being 1. During the training process, logistic regression learns the best model parameters by maximizing the likelihood function or minimizing the loss function (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis study is the first to use machine learning models to predict long-term renal function progression in patients with renal carcinoma after radical nephrectomy. This is important for early diagnosis and management of CKD. As kidney damage is irreversible, it is important to intervene and manage high-risk populations early in clinical practice. For example, the early management of diabetes is crucial. Blood sugar control can delay the progression of CKD, and most guidelines suggest a target glycated hemoglobin A1c level of ~\u0026thinsp;7.0% (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). Medication dosage and selection should also receive important attention for populations at a high risk of developing CKD. All patients at high risk for CKD should be advised to avoid nephrotoxins. While a complete list is beyond the scope of this article, there are several points worth mentioning. Chronic kidney disease Patients with CKD are advised not to routinely use nonsteroidal anti-inflammatory drugs, particularly those receiving ACE-I or ARB treatment (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Common medications that require dosage reduction include antibiotics, direct oral anticoagulants, gabapentin and pregabalin, oral hypoglycemic agents, insulin, chemotherapy drugs, and opioids (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Therefore, early dietary management is important. The KDIGO guidelines recommend that in adults at risk for CKD, reducing protein intake to below 1.3 g/kg per day may benefit from progressively restricted dietary protein, which must be balanced against the concerns of malnutrition and/or protein catabolic syndrome (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). Lowering the dietary acid load (e.g., eating more fruits and vegetables and less meat, eggs, and cheese) can also help prevent kidney damage (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eOur study has yielded an interesting conclusion. Preoperative renal function and the ratio of postoperative to preoperative creatinine levels are independent risk factors for the occurrence of CKD or deterioration of renal function in postoperative patients. The higher the preoperative renal function level, the higher is the risk of long-term deterioration of renal function. A larger creatinine ratio indicates more renal function loss after surgery than before surgery, thus increasing the risk of long-term renal function deterioration. Moreover, these two factors were the most important among all the elements in this prediction model, ranking as the top two. Olcucuoglu et al. indicated that patients who undergo radical nephrectomy might experience a more rapid deterioration of renal function than those who gradually lose kidney function (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). This may be due to the loss of nephron units in one kidney, leading to hyperfiltration and proliferation of the remaining nephron units as a compensatory hypertrophy of the remaining kidney (\u003cspan additionalcitationids=\"CR28\" citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). Initially, maintaining the GFR at a certain rate may seem beneficial, but it leads to specific structural changes, such as glomerulosclerosis and tubular atrophy. Clinically, these changes manifest as hypertension (HT), proteinuria, and GFR decline. HT and glomerulosclerosis eventually cause further nephron loss, triggering CKD development (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). This interesting finding may be counterintuitive, as it is often thought that the better the kidney function, the better is the reserve function. Therefore, this reminds us that in clinical practice, we should pay more attention to patients with initially very healthy kidney tumors, indicating that their physiological function may decline more severely after surgery.\u003c/p\u003e \u003cp\u003eHowever, our study had some limitations. First, these machine-learning models are based on single-center data for training and development, requiring further external validation to interpret the universality of these models. Second, this was a retrospective and observational study; therefore, bias was inevitable. Third, future prospective studies with larger sample sizes are necessary to further validate the potential of machine learning models to predict clinical outcomes.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study found that the better the preoperative renal function, the greater the ratio of postoperative to preoperative creatinine; the older the age, the more likely it is that renal function will deteriorate after radical nephrectomy. The logistic regression model based on patient clinical data showed the best predictive performance. It can assist clinicians in the early assessment of the risk of long-term renal function deterioration or development into CKD post-surgery, allowing for early intervention and improving patient prognosis and quality of life.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding Statement:\u003c/strong\u003e N/A\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of Interest declaration:\u0026nbsp;\u003c/strong\u003eThe authors declare that they have NO affiliations with or involvement in any organization or entity with any financial interest in the subject matter or materials discussed in this manuscript. The ethical approval number is QYFY WZLL 28808.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u003c/strong\u003e Yongchao Yan to the analysis of the results and to the writing of the manuscript. Qihang Sun and Haotian Du contributed to the design and implementation of the research, Yize Guo and Bin Li were involved in collecting data. Xinning Wang conceived the original and supervised the project.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;and consent to participate:\u003c/strong\u003e As this publication is a report that contains no identifiable content to the patient, this publication was exempt from ethical approval by the Human Research Protection Program (HRPP) and its Institutional Review Board (IRB) at the Ethics Committee of the affiliated hospital of Qingdao University.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication: N/A\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials:\u0026nbsp;\u003c/strong\u003eThe datasets used and/or analysed during the current study available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interest:\u0026nbsp;\u003c/strong\u003eThe author declare that there no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This research was supported by the Shandong Province medical health science and technology project (NO.202304051689).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAnimal Studies:\u003c/strong\u003e N/A\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRegistry and the Registration No. of the study/trial:\u003c/strong\u003e N/A\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments:\u003c/strong\u003e N/A\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eOlcucuoglu E, Tonyali S, Tastemur S, Kasap Y, Sirin ME, Gazel E, et al. Comparison of long-term kidney functions after radical nephrectomy and simple nephrectomy. PeerJ. 2019;7:e6701.\u003c/li\u003e\n\u003cli\u003eGuidi E, Cozzi M, Milani S, Spada E. Nephrectomy modifies renal angiotensin II effects in kidney donors. Am J Hypertens. 2008;21(5):592-8.\u003c/li\u003e\n\u003cli\u003eQiu Y, Zhao Q, Wang N, Yu Y, Wang R, Zhang Y, et al. Association of hypertriglyceridemic waist phenotype with renal function impairment: a cross-sectional study in a population of Chinese adults. Nutr Metab (Lond). 2020;17:63.\u003c/li\u003e\n\u003cli\u003eZabell J, Demirjian S, Lane BR, Derweesh IH, Isharwal S, Suk-Ouichai C, et al. Predictors of Long-Term Survival after Renal Cancer Surgery. J Urol. 2018;199(2):384-92.\u003c/li\u003e\n\u003cli\u003eCharles C, Ferris AH. Chronic Kidney Disease. Prim Care. 2020;47(4):585-95.\u003c/li\u003e\n\u003cli\u003eSchreuder MF. Life with one kidney. Pediatr Nephrol. 2018;33(4):595-604.\u003c/li\u003e\n\u003cli\u003eJiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, et al. Artificial intelligence in healthcare: past, present and future. Stroke and Vascular Neurology. 2017;2(4):230-43.\u003c/li\u003e\n\u003cli\u003eZabor EC, Furberg H, Lee B, Campbell S, Lane BR, Thompson RH, et al. Long-Term Renal Function Recovery following Radical Nephrectomy for Kidney Cancer: Results from a Multicenter Confirmatory Study. J Urol. 2018;199(4):921-6.\u003c/li\u003e\n\u003cli\u003eRoussel E, Laenen A, Bhindi B, De Dobbeleer A, Stichele AV, Verbeke L, et al. Predicting short- and long-term renal function following partial and radical nephrectomy. Urol Oncol. 2023;41(2):110 e1- e6.\u003c/li\u003e\n\u003cli\u003eLaguna MP. Re: Renal function after nephron-sparing surgery versus radical nephrectomy: results from EORTC randomized trial 30904. J Urol. 2014;192(2):369-70.\u003c/li\u003e\n\u003cli\u003eStoltzfus JC. Logistic regression: a brief primer. Acad Emerg Med. 2011;18(10):1099-1104. doi:10.1111/j.1553-2712.2011.01185.x\u003c/li\u003e\n\u003cli\u003eValkenborg D, Rousseau AJ, Geubbelmans M, Burzykowski T. Support vector machines. Am J Orthod Dentofacial Orthop. 2023;164(5):754-7.\u003c/li\u003e\n\u003cli\u003eRigatti SJ. Random Forest. J Insur Med. 2017;47(1):31-39. doi:10.17849/insm-47-01-31-39.1.\u003c/li\u003e\n\u003cli\u003eKoh J. Gradient boosting with extreme-value theory for wildfire prediction. Extremes (Boston). 2023;26(2):273-99.\u003c/li\u003e\n\u003cli\u003eLiao H, Zhang X, Zhao C, Chen Y, Zeng X, Li H. LightGBM: an efficient and accurate method for predicting pregnancy diseases. J Obstet Gynaecol. 2022;42(4):620-9.\u003c/li\u003e\n\u003cli\u003eOntivero-Ortega M, Lage-Castellanos A, Valente G, Goebel R, Valdes-Sosa M. Fast Gaussian Naive Bayes for searchlight classification analysis. Neuroimage. 2017;163:471-9.\u003c/li\u003e\n\u003cli\u003eArian R, Hariri A, Mehridehnavi A, Fassihi A, Ghasemi F. Protein kinase inhibitors\u0026apos; classification using K-Nearest neighbor algorithm. Comput Biol Chem. 2020;86:107269.\u003c/li\u003e\n\u003cli\u003eChen TK, Knicely DH, Grams ME. Chronic Kidney Disease Diagnosis and Management: A Review. JAMA. 2019;322(13):1294-304.\u003c/li\u003e\n\u003cli\u003eFeng X, Hong T, Liu W, Xu C, Li W, Yang B, et al. Development and validation of a machine learning model to predict the risk of lymph node metastasis in renal carcinoma. Front Endocrinol (Lausanne). 2022;13:1054358.\u003c/li\u003e\n\u003cli\u003eChen W, Zhang L, Cai G, Zhang B, Lian Z, Li J, et al. Machine learning-based multimodal MRI texture analysis for assessing renal function and fibrosis in diabetic nephropathy: a retrospective study. Front Endocrinol (Lausanne). 2023;14:1050078.\u003c/li\u003e\n\u003cli\u003eElihimas Junior UF, Couto JP, Pereira W, Barros de Oliveira Sa MP, Tenorio de Franca EE, Aguiar FC, et al. Logistic Regression Model in a Machine Learning Application to Predict Elderly Kidney Transplant Recipients with Worse Renal Function One Year after Kidney Transplant: Elderly KTbot. J Aging Res. 2020;2020:7413616.\u003c/li\u003e\n\u003cli\u003eStoltzfus JC. Logistic regression: a brief primer. Acad Emerg Med. 2011;18(10):1099-104.\u003c/li\u003e\n\u003cli\u003eInker LA, Astor BC, Fox CH, Isakova T, Lash JP, Peralta CA, et al. KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD. Am J Kidney Dis. 2014;63(5):713-35.\u003c/li\u003e\n\u003cli\u003eGuideline development g. Clinical Practice Guideline on management of patients with diabetes and chronic kidney disease stage 3b or higher (eGFR \u0026lt;45 mL/min). Nephrol Dial Transplant. 2015;30 Suppl 2:ii1-142.\u003c/li\u003e\n\u003cli\u003eAparicio M, Fouque D, Chauveau P. Effect of a very low-protein diet on long-term outcomes. Am J Kidney Dis. 2009;54(1):183.\u003c/li\u003e\n\u003cli\u003eGoraya N, Simoni J, Jo C, Wesson DE. Dietary acid reduction with fruits and vegetables or bicarbonate attenuates kidney injury in patients with a moderately reduced glomerular filtration rate due to hypertensive nephropathy. Kidney Int. 2012;81(1):86-93.\u003c/li\u003e\n\u003cli\u003eKasiske BL, Ma JZ, Louis TA, Swan SK. Long-term effects of reduced renal mass in humans. Kidney Int. 1995;48(3):814-819. doi:10.1038/ki.1995.355.\u003c/li\u003e\n\u003cli\u003eBrenner BM, Lawler EV, Mackenzie HS. The hyperfiltration theory: a paradigm shift in nephrology. Kidney Int. 1996;49(6):1774-1777. doi:10.1038/ki.1996.265.\u003c/li\u003e\n\u003cli\u003eHelal I, Fick-Brosnahan GM, Reed-Gitomer B, Schrier RW. Glomerular hyperfiltration: definitions, mechanisms and clinical implications. Nat Rev Nephrol. 2012;8(5):293-300.\u003c/li\u003e\n\u003cli\u003eSchreuder MF, Langemeijer ME, Bokenkamp A, Delemarre-Van de Waal HA, Van Wijk JA. Hypertension and microalbuminuria in children with congenital solitary kidneys. J Paediatr Child Health. 2008;44(6):363-8.\u003c/li\u003e\n\u003c/ol\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":"bmc-nephrology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bnep","sideBox":"Learn more about [BMC Nephrology](http://bmcnephrol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bnep/default.aspx","title":"BMC Nephrology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Radical nephrectomy, Chronic kidney disease, Machine Learning, Kidney cancer, Early diagnosis","lastPublishedDoi":"10.21203/rs.3.rs-5036531/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5036531/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground: Chronic Kidney Disease (CKD) is a common severe complication after radical nephrectomy in patients with renal cancer. The timely and accurate prediction of the long-term progression of renal function post-surgery is crucial for early intervention and ultimately improving patient survival rates.\u003c/p\u003e \u003cp\u003eObjective: This study aimed to establish a machine learning model to predict the likelihood of long-term renal dysfunction progression after surgery by analyzing patients\u0026rsquo; general information in depth.\u003c/p\u003e \u003cp\u003eMethods: We retrospectively collected data of eligible patients from the Affiliated Hospital of Qingdao University. The primary outcome was upgrading of the Chronic Kidney Disease stage between pre- and 3-year post-surgery. We constructed seven different machine-learning models based on Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (Lightgbm), Gaussian Naive Bayes (GaussianNB), and K-Nearest Neighbors (KNN). The performance of all predictive models was evaluated using the area under the receiver operating characteristic curve (AUC), precision-recall curves, confusion matrices, and calibration curves.\u003c/p\u003e \u003cp\u003eResults: Among 360 patients with renal cancer who underwent radical nephrectomy included in this study, 185 (51.3%) experienced an upgrade in Chronic Kidney Disease stage 3-year post-surgery. Eleven predictive variables were selected for further construction of the machine learning models. The logistic regression model provided the most accurate prediction, with the highest AUC (0.8154) and an accuracy of 0.787.\u003c/p\u003e \u003cp\u003eConclusion: The logistic regression model can more accurately predict long-term renal dysfunction progression after radical nephrectomy in patients with renal cancer.\u003c/p\u003e","manuscriptTitle":"Machine learning models predict the progression of long-term renal insufficiency in patients with renal cancer after radical nephrectomy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-03 23:33:13","doi":"10.21203/rs.3.rs-5036531/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-09-13T12:55:17+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-09-13T11:33:34+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-09-13T11:31:58+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Nephrology","date":"2024-09-05T08:22:18+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-nephrology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bnep","sideBox":"Learn more about [BMC Nephrology](http://bmcnephrol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bnep/default.aspx","title":"BMC Nephrology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5f077016-6117-40a3-ace5-dce3af845873","owner":[],"postedDate":"December 3rd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-12-23T16:01:29+00:00","versionOfRecord":{"articleIdentity":"rs-5036531","link":"https://doi.org/10.1186/s12882-024-03907-1","journal":{"identity":"bmc-nephrology","isVorOnly":false,"title":"BMC Nephrology"},"publishedOn":"2024-12-18 15:57:27","publishedOnDateReadable":"December 18th, 2024"},"versionCreatedAt":"2024-12-03 23:33:13","video":"","vorDoi":"10.1186/s12882-024-03907-1","vorDoiUrl":"https://doi.org/10.1186/s12882-024-03907-1","workflowStages":[]},"version":"v1","identity":"rs-5036531","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5036531","identity":"rs-5036531","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}