Efficacy and safety of ureteroscopy in children with lower pole renal stones – a machine learning predictive model from the EAU section of Endourology
preprint
OA: closed
CC-BY-4.0
Abstract
Abstract Introduction The rising incidence of kidney stone disease in children presents growing clinical challenges, particularly in managing lower pole (LP) calculi, which are anatomically difficult to treat. Flexible ureteroscopy with laser lithotripsy (fURSL) has emerged as a preferred minimally invasive treatment. However, surgical outcomes remain variable, especially in the paediatric LP stone cohort. This study aimed to apply machine learning (ML) techniques to predict surgical outcomes based on preoperative characteristics and identify key predictors of incomplete stone clearance. Materials and Methods A retrospective analysis was conducted on paediatric patients (< 18 years) who underwent fURSL between January 2017 and December 2021 across eight tertiary centres. From a multicentre database of 280 patients, 91 with isolated LP stones were selected. Preoperative, intraoperative, and postoperative variables were analysed. Fifteen ML models—including ensemble algorithms and a multitask neural network—were developed to predict LP stone presence and postoperative outcomes. Model performance was evaluated using accuracy, precision, recall, F1-score, and SHAP (SHapley Additive exPlanations) values for interpretability. Results LP stones were present in 32.5% of cases and were associated with older age, solitary stones, and higher stone burden. Random Forest outperformed all other models (validation accuracy: 80.95%; F1-score: 76.67%), followed by Gradient Boosting. SHAP analysis identified stone number, total stone burden, age, and operative time as top predictors. LP stones were associated with a higher rate of residual fragments (RF) and lower need for preoperative stenting or ureteral access sheath use. Infectious and bleeding complications were less frequent in the LP group. Conclusion fURSL is safe and effective in children with LP stones, though incomplete stone clearance remains a challenge. ML models demonstrated strong predictive performance and could support preoperative risk stratification. Further external validation and prospective studies are warranted to refine predictive tools for clinical use.
Full text
84,178 characters
· extracted from
preprint-html
· click to expand
Efficacy and safety of ureteroscopy in children with lower pole renal stones – a machine learning predictive model from the EAU section of Endourology | 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 Efficacy and safety of ureteroscopy in children with lower pole renal stones – a machine learning predictive model from the EAU section of Endourology Carlotta Nedbal, Vineet Gauhar, Shilpa Gite, Het Sevalia, Ratan Maurya, and 13 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7225953/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 20 Nov, 2025 Read the published version in World Journal of Urology → Version 1 posted 10 You are reading this latest preprint version Abstract Introduction The rising incidence of kidney stone disease in children presents growing clinical challenges, particularly in managing lower pole (LP) calculi, which are anatomically difficult to treat. Flexible ureteroscopy with laser lithotripsy (fURSL) has emerged as a preferred minimally invasive treatment. However, surgical outcomes remain variable, especially in the paediatric LP stone cohort. This study aimed to apply machine learning (ML) techniques to predict surgical outcomes based on preoperative characteristics and identify key predictors of incomplete stone clearance. Materials and Methods A retrospective analysis was conducted on paediatric patients (< 18 years) who underwent fURSL between January 2017 and December 2021 across eight tertiary centres. From a multicentre database of 280 patients, 91 with isolated LP stones were selected. Preoperative, intraoperative, and postoperative variables were analysed. Fifteen ML models—including ensemble algorithms and a multitask neural network—were developed to predict LP stone presence and postoperative outcomes. Model performance was evaluated using accuracy, precision, recall, F1-score, and SHAP (SHapley Additive exPlanations) values for interpretability. Results LP stones were present in 32.5% of cases and were associated with older age, solitary stones, and higher stone burden. Random Forest outperformed all other models (validation accuracy: 80.95%; F1-score: 76.67%), followed by Gradient Boosting. SHAP analysis identified stone number, total stone burden, age, and operative time as top predictors. LP stones were associated with a higher rate of residual fragments (RF) and lower need for preoperative stenting or ureteral access sheath use. Infectious and bleeding complications were less frequent in the LP group. Conclusion fURSL is safe and effective in children with LP stones, though incomplete stone clearance remains a challenge. ML models demonstrated strong predictive performance and could support preoperative risk stratification. Further external validation and prospective studies are warranted to refine predictive tools for clinical use. Paediatric urology Urolithiasis Lower pole stones Machine Learning Predictive Tools Stone Free Rates. Figures Figure 1 Figure 2 Figure 3 Introduction Kidney stone disease is an increasingly recognised condition in the paediatric population, now affecting up to 2% of children across Europe. This rising incidence is largely attributed to lifestyle changes and a growing prevalence of malnutrition-related conditions such as obesity and metabolic syndrome [ 1 ]. Early identification, effective management, and structured follow-up are crucial to prevent recurrence. However, the diagnosis and treatment of paediatric urolithiasis present unique clinical challenges. According to the current European Association of Urology (EAU) guidelines, different treatment modalities, similarly to the adult population, can be proposed and have been proved safe in children, encompassing flexible ureteroscopy with laser lithotripsy (fURSL), percutaneous nephrolithotomy (PCNL) and shockwave lithotripsy (SWL) [ 2 , 3 ]. If historically PCNL has generally been reserved for larger stones, recent advancements such as the miniaturization of endoscopic instruments, enhanced laser technologies, and the integration of suction systems have enabled successful treatment of larger calculi using less invasive techniques [ 4 ]. fURSL has emerged as a preferred minimally invasive approach, with good safety and efficacy profile. Some debates still remain on the best approach with complex stones, and in particular for lower pole (LP) calculi. Despite the known advantages of fURSL, with lower complication rates and wider applicability compared to PCNL [ 5 ], treatment outcomes remain highly variable, influenced by both anatomical complexities and preoperative factors, areas that are the focus of ongoing research [ 6 ]. Among stone locations, LP calculi remain the most technically challenging to manage via fURSL. This difficulty arises from unfavourable anatomy, particularly the steep infundibular pelvic angle (IPA), which impairs access and hinders complete fragmentation. As a result, paediatric patients with lower pole stones often face longer, more complex procedures and lower stone-free rates (SFR) [ 7 ]. Although technological progress has improved procedural capabilities, real-world data suggest that achieving complete stone clearance remains difficult [ 8 , 9 ]. To better predict surgical outcomes, nomograms have been developed to estimate SFR in paediatric patients, incorporating variables such as stone burden, location, infection status, and individual anatomy [ 10 ]. Machine learning (ML) offers significant potential in this context, enabling the analysis of complex, multidimensional data to identify predictive patterns that traditional statistical approaches may overlook [ 11 , 12 ]. Complementing this, Explainable Artificial Intelligence (XAI) tools are increasingly used to improve transparency, interpretability, and clinical trust in ML-driven predictions, thereby supporting more informed and safer decision-making. The present study retrospectively evaluated paediatric patients undergoing fURSL for isolated lower pole stones and applied ML techniques to predict surgical outcomes based on preoperative characteristics. By identifying key predictors, our aim is to facilitate individualized risk stratification and enhance surgical planning in this vulnerable population. Materials and Methods Patient Selection, Data Collection, and Operative Protocols A retrospective review was performed on paediatric patients (< 18 years) who underwent fURSL for urolithiasis between January 2017 and December 2021 across eight high-volume tertiary centres. Data were extracted from a previously published multicentric database comprising more than 300 patients [ 7 ]. For this study, patients presenting with isolated lower pole stones were selected, resulting in a final cohort of 91 patients, and compared with the non-LP cohort. Demographic and clinical variables collected included age, sex, comorbidities, stone burden (single vs. multiple), and maximum stone diameter. The presence of known metabolic or genetic disorders such as renal tubular acidosis (RTA), hypercalciuria, cystinuria, or hyperoxaluria/hypocitraturia was recorded. Additional variables included presenting symptoms, history of stone recurrence, and prior interventions. Preoperative imaging modalities followed institutional protocols and included non-contrast or contrast-enhanced computed tomography (CT) and/or dedicated renal ultrasonography (USS). Patients were excluded if they had non-urolithiasis diagnoses (e.g., upper tract malignancies), anatomical renal anomalies, stones in non-lower-pole locations, were aged ≥ 18 years, or lacked consent for data usage. Informed consent was obtained from all patients and/or legal guardians following specialist counselling. A sterile urine culture was required prior to surgery. Intraoperative data included operative time, use of a ureteral access sheath (UAS), type of ureteroscope (reusable or single-use), laser type, and postoperative ureteral stenting. Postoperative variables included haematuria, fever, sepsis, presence and number of residual fragments (RF), and requirement for reintervention. Stone-free status was defined as no RF or RF < 2 mm on low-dose CT-KUB or USS at 3-month follow-up. Machine Learning Model Development and Analysis Following data preprocessing, which included removal of irrelevant symbols and imputation of missing categorical data using mode values, statistical evaluations were conducted. These included correlation analyses, Variance Inflation Factor (ViF) assessments, and logistic regression across four predefined outcome prediction tasks. A total of fifteen ML models were developed and individually trained to predict clinical outcomes using preoperative variables. The models encompassed multiple algorithmic categories: Probabilistic Methods: Naive Bayes, Logistic Regression Instance-Based Learning: K-Nearest Neighbors (KNN) Discriminant Analysis: Linear (LDA) and Quadratic Discriminant Analysis (QDA) Support Vector Machines (SVM): Polynomial and RBF kernels Tree-Based Algorithms: Decision Tree, Random Forest, Extra Trees, Bagging Classifier, Gradient Boosting, CatBoost, XGBoost, AdaBoost Additionally, a multitask Artificial Neural Network (ANN) was developed to simultaneously predict all target outcomes. The ANN architecture consisted of shared hidden layers for feature extraction and individual task-specific output layers. Rectified Linear Unit (ReLU) activations were employed in hidden layers, and sigmoid activations were used in the output layers. Model performance was evaluated using classification reports and confusion matrices, including accuracy, precision, recall, and F1-score. To improve interpretability of complex models (e.g., ANN), explainable AI (XAI) techniques, specifically SHAP (SHapley Additive exPlanations) values were employed to highlight the most influential features in model predictions. Model Evaluation and Performance Metrics Bar charts were generated to compare model validation accuracy across various feature subsets (e.g., patient demographics, stone characteristics, metabolic parameters). Validation accuracy represented model generalizability on unseen data. Autologger was used to ensure automatic tracking of model parameters, enabling transparent and reproducible evaluation. Feature groups with higher validation accuracy were interpreted as having greater predictive value. Additional bar plots presented training accuracies and loss metrics to assess learning behaviour across different models and feature sets. Training accuracy served to evaluate how well models fit the training data. These metrics helped identify potential overfitting when training accuracy was high, but validation performance was suboptimal. Autologger also recorded training time and loss functions to compare algorithmic efficiency and convergence. Confusion matrices were employed to visualise classification performance by mapping predicted versus actual outcomes for each model, aiding in the identification of false positives and negatives. Each model's classification report included key performance indicators: Precision: proportion of positive predictions that were correct; Recall: proportion of actual positives correctly identified; F1-score: harmonic mean of precision and recall; Accuracy: overall proportion of correct predictions Results Descriptive and Statistical Findings A total of 280 paediatric patients were analysed, of whom 91 (32.5%) had stones located exclusively in the LP. The dataset comprised 29 variables, encompassing demographic, clinical, imaging, and procedural data. Imputation was applied for 499 missing values (numerical variables via median; categorical via mode). Comparing LP stones with the control cohort, patients affected by LP stones appeared to be slightly older, and presenting with higher total stone burden but not larger single-stone diameter (Fig. 1 ). Correlation analysis Correlation analysis revealed that no single variable exhibited strong linear association with LP stone presence (Table 1 ). The most positively correlated features were residual fragments (RF; r = + 0.106), total stone burden (r = + 0.101), and age (r = + 0.045). These results suggest that LP stones are more common in slightly older patients and in those with greater stone volume but are also related to a higher risk of incomplete clearance and RF, compared to the general population. In contrast, the presence of multiple stones (r = − 0.154), haematuria (r = − 0.138), pre-stenting (r = − 0.135), and use of a ureteral access sheath (UAS; r = − 0.128) were negatively associated with LP stones. This implies that LP stones tend to be solitary, less symptomatic, and can be often managed without UAS or prior stenting (Fig. 2 ). There was no significant difference in stone density between the two groups, due to the fact that HU are mostly determined by stone composition and not stone location. Postoperative complications did not show significant patterns linked with the presence of LP stones. Infectious complications like fever (r = -0.091) and sepsis (r = -0.072) showed only mild negative correlation with LP stones, and similar results were found for the presence of positive preoperative MSU (r = -0.054). Table 1 Main results of correlation analysis, divided in negative and positive correlations (with relative r values). Positive correlations Negative correlations Feature Correlation Feature Correlation RF + 0.106 Stone number −0.154 Total stone burden (mm) + 0.101 Haematuria −0.138 Reintervention + 0.047 Presented −0.135 Age (years) + 0.045 UAS used −0.128 HolmLP + 0.044 Laser time (minutes) −0.076 Reusable scope + 0.040 Fever −0.091 Known genetic disorders + 0.038 Sepsis −0.072 Postoperative stent + 0.034 Positive MSU −0.054 Normal kidney anatomy + 0.012 Multiple fragments −0.032 A hierarchical clustering analysis identified feature groups with tight interdependencies, notably between UAS use and sheath size (VIF > 70), warranting the removal of one to reduce multicollinearity. Pairwise comparisons confirmed that LP stones were more frequently associated with single stones, lower haematuria rates, and reduced use of UAS and larger sheaths. A detailed overview of the correlation and ML analysis in available in the supplementary materials section. Machine Learning Model Performance Fifteen machine learning classifiers were trained using preoperative variables to predict the presence of LP stones. Ensemble-based models significantly outperformed individual classifiers. Random Forest emerged as the top-performing model with a validation accuracy of 80.95%, precision of 80.63%, recall of 75.00%, and F1-score of 76.67%, reflecting excellent balance between sensitivity and specificity. Gradient Boosting demonstrated similarly robust generalization with 76.19% validation accuracy and balanced metric profiles (training accuracy: 97.06%). Extra Trees Classifier reached 73.81% validation accuracy but showed signs of overfitting due to perfect training performance. Other strong models included CatBoost and Bagging Classifier, both achieving 71.43% validation accuracy. In contrast, Quadratic Discriminant Analysis (QDA) and Naïve Bayes underperformed (validation accuracy ~ 40–50%), while Support Vector Machine variants displayed modest accuracy (~ 66.67%) but low F1-scores, indicating poor class boundary definition. Feature Importance and Explainability Explainable AI techniques, including SHAP (SHapley Additive exPlanations) and decision tree visualizations, provided interpretability of model predictions. Top predictors of LP stone presence included: Stone number (single vs. multiple); Total operative time; Maximum stone diameter; Total stone burden; Age; Residual Fragments; UAS usage; and Presentation status (pre-stenting). SHAP summary plots indicated that multiple stones most consistently increased the model’s likelihood of predicting LP stones, and this stone location was linked to shorter laser times and higher risk of RF. Conversely, features such as female sex, avoiding UAS use, and shorter operative time contributed negatively, reducing LP stone probability in specific predictions. Decision tree visualizations confirmed that stone number was the initial node in model splitting, with subsequent branches involving RF, stone diameter, and fragmentation patterns. This hierarchy supported the clinical assumption that solitary stones and minimal instrumentation are more typical in LP stone cases. Prediction-level explanations (via SHAP waterfall and force plots) confirmed that individual outcomes were strongly driven by procedural details and stone configuration, rather than isolated demographic or biochemical features. Summary of Predictive Accuracy Overall, ensemble models, especially Random Forest and Gradient Boosting demonstrated superior performance in identifying LP stones, with validation accuracies exceeding 75%. While individual predictors showed weak linear correlation with LP stone presence, their collective interaction in nonlinear models yielded clinically meaningful predictive power. These results highlight the value of ML in supporting diagnostic decision-making for paediatric urolithiasis. Discussion The ML model trained well on the retrospective database, providing good accuracy in prediction of LP related features and outcomes. To our knowledge, this is currently the largest ML analysis on paediatric patients undergoing fURSL, and the only one to investigate the role of LP stones in probability of stone clearance and risk of complications. As suspected, we found an increased risk of incomplete clearance in this population, alongside the need for multiple intervention in light of both the risk of residual fragments and recurrence of stones. Interestingly, infectious complications appear to be slightly less frequent in children with LP stones compared to other stone locations, and this is reflected by the lower rates of positive urine cultures and patients needing extended preoperative antibiotics. Postoperative haematuria was also less frequently encountered in this population group, despite the well-known challenges that a stiff infundibular angle can pose while treating LP stones [ 13 ]. In our opinion, these findings might somehow be related, given the increased risk of contact-bleeding and mucosal abrasion during laser treatment when dealing with inflamed tissues, characteristics of patients with recurrent urinary tract infections or infected stones. Based on our analysis, preoperative presentations and stone characteristics can differ in children affected by LP and non-LP stones. Older children and adolescents tend to have larger stones located in the LP, but they often present with single stones that are not obstructive and therefore require preoperative stenting less frequently. These findings may relate with renal anatomy, with gravity and infundibular angle playing a role in the larger growth of urolithiasis in LP stones before they get symptomatic and prompt interventions. Similarly, we found a moderate correlation between LP stones and known genetic disorders, with increased risk of urine stasis and stone formation. This link appears to be stronger that the one with anatomical variants of the kidney. Previous studies, not performed with the innovative aid of ML analysis, found an increased risk for incomplete clearance in children with LP stones. These results are in line with our analysis, with same risk of operative and postoperative complications reported in the two subgroups [ 14 ]. As also described by Sen and colleagues, LP stones can be difficult to completely remove, but RF in these cases can frequently be asymptomatic and more rarely require reintervention [ 15 ]. Nevertheless, thanks to technological advancements and miniaturisation of fURSL instruments, ureteroscopy can now considered a first line treatment in children with LP stones, and a two-step procedure can still be offered as complication rates and impact on children quality of life appears to be improved compared to PCNL [ 16 ]. Previous systematic review looking at the role of fURSL for paediatric renal stones recommends that these procedures are done by experienced surgeons [17]. Our study represents one of the first ML-based analysis on a paediatric population undergoing fURSL for urolithiasis, and the first to describe findings on a large cohort with LP stones. Despite the retrospective nature, that comes with possible biases of missing or incomplete data, the analysis and ML training resulted in good performance and identification of predictive factors and outcomes. This multicentric database collected a large number of paediatric cases and reflects the variety of real-word practice. We believe that this represents a strength of the study, and with further external validation and prospective analysis a predictive model might be developed for this fragile subgroup with increased risk of incomplete stone clearance. Perhaps future studies should also look at the cost of fURSL procedures and adaption of ML techniques in outcome prediction for these procedures [18,19]. Conclusion Children with LP stones can be safely and effectively treated with fURSL, but a higher risk of incomplete stone clearance needs to be acknowledged. Obstructive stones are less frequent in patients with LP stones, and there is a reduced need for preoperative stent insertion. Complications rates appear to be similar to the general paediatric population. ML models demonstrated strong predictive performance and could support preoperative risk stratification, but further external validation and implementation with prospective data are warranted to refine predictive tools for clinical use. Declarations Consent to Participate: all patients included in the study signed an informed consent for data collection and research analysis. Ethics Approval: given the anonymous and retrospective nature of the study, no ethical approval was needed according to our Internal Review Board (IRB). Funding Declaration: no Funding. Human Ethics and Consent to Participate declarations: not applicable. Author Contribution CN: writing, editing, project conceptualisationVG, SKKY, EJL: data collection, project conceptualisation, editingSG, HT, RM, PJ, KK: data analysis, results, figuresAG, FA, FP, YT, AS, BS, HYL, NN: project conceptualisation, editingBKS: project conception, editing, supervisionAll authors reviewed the manuscript Acknowledgement DeclarationsConsent to Participate: all patients included in the study signed an informed consent for data collection and research analysis.Ethics Approval: given the anonymous and retrospective nature of the study, no ethical approval was needed according to our Internal Review Board (IRB).Funding Declaration: no Funding.Human Ethics and Consent to Participate declarations: not applicable. References Jobs K, Rakowska M, Paturej A (2018) Urolithiasis in the pediatric population - current opinion on epidemiology, patophysiology, diagnostic evaluation and treatment. Dev Period Med 22:201–208. https://doi.org/10.34763/devperiodmed.20182202.201208 Skolarikos A, Neisius A, Petřík A et al (2022) EAU Guidelines on Urolithiasis. In: EAU guidelines. Edn. presented at the EAU Annual congress Amsterdam. https://uroweb.org/guidelines/urolithiasis/chapter/introduction Tsai S-H, Chung H-J, Tseng P-T et al (2020) Comparison of the efficacy and safety of shockwave lithotripsy, retrograde intrarenal surgery, percutaneous nephrolithotomy, and minimally invasive percutaneous nephrolithotomy for lower-pole renal stones. Medicine 99:e19403. https://doi.org/10.1097/MD.0000000000019403 Frascheri MF, Contreras P, Enikeev D et al (2025) Small tools, big impact and redefining endourology: the paradigm shift of ureteroscopy for large stones and percutaneous nephrolithotomy for small stones – a literature review. Curr Opin Urol 35:347–353. https://doi.org/10.1097/MOU.0000000000001283 Kandemir A, Guven S, Balasar M et al (2017) A prospective randomized comparison of micropercutaneous nephrolithotomy (Microperc) and retrograde intrarenal surgery (RIRS) for the management of lower pole kidney stones. World J Urol 35:1771–1776. https://doi.org/10.1007/s00345-017-2058-9 Alma E, Altunkol A, Anıl H et al (2025) Comparision of scoring systems in the success of retrograde intrarenal surgery and factors affecting the presence of residual fragments in pediatric patients. J Pediatr Urol. https://doi.org/10.1016/j.jpurol.2025.06.011 Arikan Y (2024) In the Treatment of Lower Pole Kidney Stones Between 1–2 cm in Children, Which is the Best Approach? Retrograde Intrarenal Surgery or Mini Percutaneous Nephrolithotomy. SiSli Etfal Hastanesi Tip Bulteni /. Med Bull Sisli Hosp 332–338. https://doi.org/10.14744/semb.2024.49225 Kim HJ, Daignault-Newton S, DiBianco JM et al (2023) Real-world Practice Stone-free Rates After Ureteroscopy: Variation and Outcomes in a Surgical Collaborative. Eur Urol Focus 9:773–780. https://doi.org/10.1016/j.euf.2023.03.010 Mille E, El-Khoury E, Haddad M et al (2023) Comparison of single-use flexible ureteroscopes with a reusable ureteroscope for the management of paediatric urolithiasis. J Pediatr Urol 19:248e1–248e6. https://doi.org/10.1016/j.jpurol.2023.01.009 Rosière NI, Ruiz J, López Imizcoz F et al (2025) Predictive variables for successful retrograde intrarenal surgery outcomes in pediatric patients: Development of nomograms. J Pediatr Urol 21:775–776. https://doi.org/10.1016/j.jpurol.2025.01.012 Nedbal C, Adithya S, Gite S et al (2024) A Machine Learning Predictive Model for Ureteroscopy Lasertripsy Outcomes in a Pediatric Population—Results from a Large Endourology Tertiary Center. J Endourol. https://doi.org/10.1089/end.2024.0120 Dong J, Feng T, Thapa-Chhetry B et al (2021) Machine learning model for early prediction of acute kidney injury (AKI) in pediatric critical care. Crit Care 25:288. https://doi.org/10.1186/s13054-021-03724-0 Fernandez N, Ellison JS, Wang Z et al (2024) Surgeon, and Institution Characteristics Associated Surgical Preferences in the Pediatric KIDney Stone Care Improvement Network. Urology 187:64–70. https://doi.org/10.1016/j.urology.2024.02.040 Unal U, Deniz ME, Ortoglu F et al (2022) What is the best approach for pediatric kidney stones of moderate-sized between shock wave lithotripsy, ultramini percutaneous nephrolithotomy and retrograde intrarenal surgery? Pediatr Surg Int 38:1643–1648. https://doi.org/10.1007/s00383-022-05203-1 Sen H, Baydilli N, Ozturk M et al (2024) Factors effecting the success of retrograde intrarenal surgery in pediatric patients with renal stones: The experience of two tertiary centres with 368 renal units. J Pediatr Urol 20. https://doi.org/10.1016/j.jpurol.2024.01.004 . :403.e1-403.e9 Mosquera L, Pietropaolo A, Madarriaga YQ et al (2021) Is Flexible Ureteroscopy and Laser Lithotripsy the New Gold Standard for Pediatric Lower Pole Stones? Outcomes from Two Large European Tertiary Pediatric Endourology Centers. J Endourol 35:1479–1482. https://doi.org/10.1089/end.2020.1123 Additional Declarations No competing interests reported. Supplementary Files SUPPLEMENTARYMATERIALS.docx Cite Share Download PDF Status: Published Journal Publication published 20 Nov, 2025 Read the published version in World Journal of Urology → Version 1 posted Editorial decision: Revision requested 20 Oct, 2025 Reviews received at journal 23 Aug, 2025 Reviews received at journal 12 Aug, 2025 Reviewers agreed at journal 08 Aug, 2025 Reviewers agreed at journal 07 Aug, 2025 Reviewers agreed at journal 06 Aug, 2025 Reviewers invited by journal 05 Aug, 2025 Editor assigned by journal 29 Jul, 2025 Submission checks completed at journal 29 Jul, 2025 First submitted to journal 27 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7225953","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":497393398,"identity":"c1ca2e4b-a15f-4bc4-9048-7cd100e03ea2","order_by":0,"name":"Carlotta Nedbal","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/UlEQVRIiWNgGAWjYHACxgNQhhkDQ4ENSKDxAG7VEHCAIcEAqsUgDaSlgSQth6EieAB/+9kHBz7++CNvzt687cEPg/N2a9sPA22psYnGpUXiTLrBwRkJBoY7e46VG/YY3E7ediYRqOVYWm4DDi0GDGkMh3kSDBg33Mgxk+ABajE7ANTC2HAYtxb+Z2At9hvuvzGT/GNwLtns/EMCWiQgtiRuuMFjJs1jcMDO7AYBWyRuPGM4OCPNOHnDmbRyYxmD5ASzG0BbEvD4hb8/jfHBBxs52w3HD297+KbCzt7sfPrDBx9qbHBqwQCJYJUJxCoHAXtSFI+CUTAKRsHIAACawWXN79CMoQAAAABJRU5ErkJggg==","orcid":"","institution":"Polytechnic University Le Marche","correspondingAuthor":true,"prefix":"","firstName":"Carlotta","middleName":"","lastName":"Nedbal","suffix":""},{"id":497393399,"identity":"d619640a-239f-40fd-b8ec-cdf0081b5d9e","order_by":1,"name":"Vineet Gauhar","email":"","orcid":"","institution":"University Hospital Southampton NHS Foundation Trust","correspondingAuthor":false,"prefix":"","firstName":"Vineet","middleName":"","lastName":"Gauhar","suffix":""},{"id":497393400,"identity":"7f04f207-5476-4b19-a5b7-94dae9e24231","order_by":2,"name":"Shilpa Gite","email":"","orcid":"","institution":"Symbiosis Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Shilpa","middleName":"","lastName":"Gite","suffix":""},{"id":497393401,"identity":"d2a96308-0d90-47cb-9a7e-4c7c6662ea9e","order_by":3,"name":"Het Sevalia","email":"","orcid":"","institution":"Symbiosis Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Het","middleName":"","lastName":"Sevalia","suffix":""},{"id":497393402,"identity":"dafdf0e9-5e6f-4758-999b-c695de8a8535","order_by":4,"name":"Ratan Maurya","email":"","orcid":"","institution":"Symbiosis Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Ratan","middleName":"","lastName":"Maurya","suffix":""},{"id":497393403,"identity":"5cd87c4e-46df-499d-bee4-6fc559d7d49e","order_by":5,"name":"Prisha Jaiswal","email":"","orcid":"","institution":"Symbiosis Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Prisha","middleName":"","lastName":"Jaiswal","suffix":""},{"id":497393404,"identity":"5079412c-f3f2-42cd-8bfc-35e393030eb6","order_by":6,"name":"Khushi Kashyap","email":"","orcid":"","institution":"Symbiosis Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Khushi","middleName":"","lastName":"Kashyap","suffix":""},{"id":497393405,"identity":"37e10d1b-c55d-4c7b-82c9-45d9911fe01c","order_by":7,"name":"Andrea Gregori","email":"","orcid":"","institution":"IRCSS San Gerardo","correspondingAuthor":false,"prefix":"","firstName":"Andrea","middleName":"","lastName":"Gregori","suffix":""},{"id":497393406,"identity":"e2d6fe15-052f-4d9b-a513-e4bf5c0b0fca","order_by":8,"name":"Francesco Antomarchi","email":"","orcid":"","institution":"IRCSS San Gerardo","correspondingAuthor":false,"prefix":"","firstName":"Francesco","middleName":"","lastName":"Antomarchi","suffix":""},{"id":497393407,"identity":"8e7aaed2-e45f-4236-ac14-9fc1b25dc598","order_by":9,"name":"Frédéric Panthier","email":"","orcid":"","institution":"Sorbonne University, Tenon Hospital","correspondingAuthor":false,"prefix":"","firstName":"Frédéric","middleName":"","lastName":"Panthier","suffix":""},{"id":497393408,"identity":"f18827cc-b47d-46d7-b57c-a96cfd340676","order_by":10,"name":"Yiloren Tanidir","email":"","orcid":"","institution":"Marmara University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yiloren","middleName":"","lastName":"Tanidir","suffix":""},{"id":497393409,"identity":"01598c05-50a2-4fce-ae5c-d95bfd819cca","order_by":11,"name":"Abhishek Singh","email":"","orcid":"","institution":"Muljibhai Patel Urological Hospital","correspondingAuthor":false,"prefix":"","firstName":"Abhishek","middleName":"","lastName":"Singh","suffix":""},{"id":497393410,"identity":"244ab70f-b39d-4ff4-becc-c798f9a196c4","order_by":12,"name":"Boyke Soebhali","email":"","orcid":"","institution":"Muliawarman University","correspondingAuthor":false,"prefix":"","firstName":"Boyke","middleName":"","lastName":"Soebhali","suffix":""},{"id":497393411,"identity":"c2cb1fd4-c58f-4063-ab75-479e7c5c0b68","order_by":13,"name":"Hsiang Ying Lee","email":"","orcid":"","institution":"Kaohsiung Medical University","correspondingAuthor":false,"prefix":"","firstName":"Hsiang","middleName":"Ying","lastName":"Lee","suffix":""},{"id":497393412,"identity":"f0193d89-593a-4736-972b-026d17871e95","order_by":14,"name":"Steffi Kar Kei Yuen","email":"","orcid":"","institution":"Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Steffi","middleName":"Kar Kei","lastName":"Yuen","suffix":""},{"id":497393413,"identity":"8649c223-3244-41ff-9724-7e1f3e693589","order_by":15,"name":"Ee Jean Lim","email":"","orcid":"","institution":"Singapore General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Ee","middleName":"Jean","lastName":"Lim","suffix":""},{"id":497393414,"identity":"8d1e3434-f02a-424f-b9a2-0cf83e27e91b","order_by":16,"name":"Nitesh Naik","email":"","orcid":"","institution":"Manipal Academy of Higher Education","correspondingAuthor":false,"prefix":"","firstName":"Nitesh","middleName":"","lastName":"Naik","suffix":""},{"id":497393415,"identity":"c24e0119-7508-4cad-88df-8287563aa5f9","order_by":17,"name":"Bhaskar Kumar Somani","email":"","orcid":"","institution":"University Hospital Southampton NHS Foundation Trust","correspondingAuthor":false,"prefix":"","firstName":"Bhaskar","middleName":"Kumar","lastName":"Somani","suffix":""}],"badges":[],"createdAt":"2025-07-27 11:38:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7225953/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7225953/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00345-025-06095-1","type":"published","date":"2025-11-20T15:58:55+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":88777260,"identity":"2e9c69d4-bbfd-4562-abae-009315b805ad","added_by":"auto","created_at":"2025-08-11 10:11:26","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":139293,"visible":true,"origin":"","legend":"\u003cp\u003ePairwise Feature Comparison By LP Stone\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7225953/v1/f94f41ed6b8714fe08816f11.jpg"},{"id":88774800,"identity":"ae2f5c25-25f7-461d-8766-7a7fecfa6573","added_by":"auto","created_at":"2025-08-11 10:03:26","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":92523,"visible":true,"origin":"","legend":"\u003cp\u003eThe radar plot visualizes the normalized mean values of the top predictive features that distinguish patients with LP stones (red) from those with non-LP stones (blue). This multi-feature representation enables side-by-side comparison of clinical patterns, aiding in the identification of relevant trends associated with LP stone presence.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7225953/v1/4a384e5dac7241bb787d1d48.jpg"},{"id":88777261,"identity":"f622cb52-c6d6-4f6d-97af-a041632531bd","added_by":"auto","created_at":"2025-08-11 10:11:26","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":112896,"visible":true,"origin":"","legend":"\u003cp\u003eExplainable AI for LP stones. From the top: Feature Importance, SHAP Waterfall Plot (Detailed Feature Decomposition for One Prediction), SHAP Force Plot (Prediction-Level Impact Visualization).\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7225953/v1/fd68c691cdffceabaaf912c8.jpg"},{"id":96650416,"identity":"7a9e73fe-6c38-407f-996f-3927b55c3585","added_by":"auto","created_at":"2025-11-24 16:12:06","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":919059,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7225953/v1/b2a98d0c-258e-40f8-8030-bd0ff4f91c02.pdf"},{"id":88774809,"identity":"f62d4ea3-9331-4a82-89b3-4d3f69e39d20","added_by":"auto","created_at":"2025-08-11 10:03:26","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":2561460,"visible":true,"origin":"","legend":"","description":"","filename":"SUPPLEMENTARYMATERIALS.docx","url":"https://assets-eu.researchsquare.com/files/rs-7225953/v1/7118ebc3001e7e06f726c3b6.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Efficacy and safety of ureteroscopy in children with lower pole renal stones – a machine learning predictive model from the EAU section of Endourology","fulltext":[{"header":"Introduction","content":"\u003cp\u003eKidney stone disease is an increasingly recognised condition in the paediatric population, now affecting up to 2% of children across Europe. This rising incidence is largely attributed to lifestyle changes and a growing prevalence of malnutrition-related conditions such as obesity and metabolic syndrome [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Early identification, effective management, and structured follow-up are crucial to prevent recurrence. However, the diagnosis and treatment of paediatric urolithiasis present unique clinical challenges.\u003c/p\u003e\u003cp\u003eAccording to the current European Association of Urology (EAU) guidelines, different treatment modalities, similarly to the adult population, can be proposed and have been proved safe in children, encompassing flexible ureteroscopy with laser lithotripsy (fURSL), percutaneous nephrolithotomy (PCNL) and shockwave lithotripsy (SWL) [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. If historically PCNL has generally been reserved for larger stones, recent advancements such as the miniaturization of endoscopic instruments, enhanced laser technologies, and the integration of suction systems have enabled successful treatment of larger calculi using less invasive techniques [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e\u003cp\u003efURSL has emerged as a preferred minimally invasive approach, with good safety and efficacy profile. Some debates still remain on the best approach with complex stones, and in particular for lower pole (LP) calculi. Despite the known advantages of fURSL, with lower complication rates and wider applicability compared to PCNL [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], treatment outcomes remain highly variable, influenced by both anatomical complexities and preoperative factors, areas that are the focus of ongoing research [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Among stone locations, LP calculi remain the most technically challenging to manage via fURSL. This difficulty arises from unfavourable anatomy, particularly the steep infundibular pelvic angle (IPA), which impairs access and hinders complete fragmentation. As a result, paediatric patients with lower pole stones often face longer, more complex procedures and lower stone-free rates (SFR) [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eAlthough technological progress has improved procedural capabilities, real-world data suggest that achieving complete stone clearance remains difficult [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. To better predict surgical outcomes, nomograms have been developed to estimate SFR in paediatric patients, incorporating variables such as stone burden, location, infection status, and individual anatomy [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eMachine learning (ML) offers significant potential in this context, enabling the analysis of complex, multidimensional data to identify predictive patterns that traditional statistical approaches may overlook [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Complementing this, Explainable Artificial Intelligence (XAI) tools are increasingly used to improve transparency, interpretability, and clinical trust in ML-driven predictions, thereby supporting more informed and safer decision-making.\u003c/p\u003e\u003cp\u003eThe present study retrospectively evaluated paediatric patients undergoing fURSL for isolated lower pole stones and applied ML techniques to predict surgical outcomes based on preoperative characteristics. By identifying key predictors, our aim is to facilitate individualized risk stratification and enhance surgical planning in this vulnerable population.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cem\u003ePatient Selection, Data Collection, and Operative Protocols\u003c/em\u003e\u003c/p\u003e\u003cp\u003eA retrospective review was performed on paediatric patients (\u0026lt;\u0026thinsp;18 years) who underwent fURSL for urolithiasis between January 2017 and December 2021 across eight high-volume tertiary centres. Data were extracted from a previously published multicentric database comprising more than 300 patients [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. For this study, patients presenting with isolated lower pole stones were selected, resulting in a final cohort of 91 patients, and compared with the non-LP cohort.\u003c/p\u003e\u003cp\u003eDemographic and clinical variables collected included age, sex, comorbidities, stone burden (single vs. multiple), and maximum stone diameter. The presence of known metabolic or genetic disorders such as renal tubular acidosis (RTA), hypercalciuria, cystinuria, or hyperoxaluria/hypocitraturia was recorded. Additional variables included presenting symptoms, history of stone recurrence, and prior interventions.\u003c/p\u003e\u003cp\u003ePreoperative imaging modalities followed institutional protocols and included non-contrast or contrast-enhanced computed tomography (CT) and/or dedicated renal ultrasonography (USS). Patients were excluded if they had non-urolithiasis diagnoses (e.g., upper tract malignancies), anatomical renal anomalies, stones in non-lower-pole locations, were aged\u0026thinsp;\u0026ge;\u0026thinsp;18 years, or lacked consent for data usage. Informed consent was obtained from all patients and/or legal guardians following specialist counselling. A sterile urine culture was required prior to surgery.\u003c/p\u003e\u003cp\u003eIntraoperative data included operative time, use of a ureteral access sheath (UAS), type of ureteroscope (reusable or single-use), laser type, and postoperative ureteral stenting. Postoperative variables included haematuria, fever, sepsis, presence and number of residual fragments (RF), and requirement for reintervention. Stone-free status was defined as no RF or RF\u0026thinsp;\u0026lt;\u0026thinsp;2 mm on low-dose CT-KUB or USS at 3-month follow-up.\u003c/p\u003e\u003cp\u003e\u003cem\u003eMachine Learning Model Development and Analysis\u003c/em\u003e\u003c/p\u003e\u003cp\u003eFollowing data preprocessing, which included removal of irrelevant symbols and imputation of missing categorical data using mode values, statistical evaluations were conducted. These included correlation analyses, Variance Inflation Factor (ViF) assessments, and logistic regression across four predefined outcome prediction tasks.\u003c/p\u003e\u003cp\u003eA total of fifteen ML models were developed and individually trained to predict clinical outcomes using preoperative variables. The models encompassed multiple algorithmic categories:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eProbabilistic Methods: Naive Bayes, Logistic Regression\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eInstance-Based Learning: K-Nearest Neighbors (KNN)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eDiscriminant Analysis: Linear (LDA) and Quadratic Discriminant Analysis (QDA)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eSupport Vector Machines (SVM): Polynomial and RBF kernels\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eTree-Based Algorithms: Decision Tree, Random Forest, Extra Trees, Bagging Classifier, Gradient Boosting, CatBoost, XGBoost, AdaBoost\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eAdditionally, a multitask Artificial Neural Network (ANN) was developed to simultaneously predict all target outcomes. The ANN architecture consisted of shared hidden layers for feature extraction and individual task-specific output layers. Rectified Linear Unit (ReLU) activations were employed in hidden layers, and sigmoid activations were used in the output layers.\u003c/p\u003e\u003cp\u003eModel performance was evaluated using classification reports and confusion matrices, including accuracy, precision, recall, and F1-score. To improve interpretability of complex models (e.g., ANN), explainable AI (XAI) techniques, specifically SHAP (SHapley Additive exPlanations) values were employed to highlight the most influential features in model predictions.\u003c/p\u003e\u003cp\u003e\u003cem\u003eModel Evaluation and Performance Metrics\u003c/em\u003e\u003c/p\u003e\u003cp\u003eBar charts were generated to compare model validation accuracy across various feature subsets (e.g., patient demographics, stone characteristics, metabolic parameters). Validation accuracy represented model generalizability on unseen data. Autologger was used to ensure automatic tracking of model parameters, enabling transparent and reproducible evaluation. Feature groups with higher validation accuracy were interpreted as having greater predictive value.\u003c/p\u003e\u003cp\u003eAdditional bar plots presented training accuracies and loss metrics to assess learning behaviour across different models and feature sets. Training accuracy served to evaluate how well models fit the training data. These metrics helped identify potential overfitting when training accuracy was high, but validation performance was suboptimal. Autologger also recorded training time and loss functions to compare algorithmic efficiency and convergence.\u003c/p\u003e\u003cp\u003eConfusion matrices were employed to visualise classification performance by mapping predicted versus actual outcomes for each model, aiding in the identification of false positives and negatives. Each model's classification report included key performance indicators: Precision: proportion of positive predictions that were correct; Recall: proportion of actual positives correctly identified; F1-score: harmonic mean of precision and recall; Accuracy: overall proportion of correct predictions\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cem\u003eDescriptive and Statistical Findings\u003c/em\u003e\u003c/p\u003e\u003cp\u003eA total of 280 paediatric patients were analysed, of whom 91 (32.5%) had stones located exclusively in the LP. The dataset comprised 29 variables, encompassing demographic, clinical, imaging, and procedural data. Imputation was applied for 499 missing values (numerical variables via median; categorical via mode). Comparing LP stones with the control cohort, patients affected by LP stones appeared to be slightly older, and presenting with higher total stone burden but not larger single-stone diameter (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eCorrelation analysis\u003c/em\u003e\u003c/p\u003e\u003cp\u003eCorrelation analysis revealed that no single variable exhibited strong linear association with LP stone presence (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The most positively correlated features were residual fragments (RF; r\u0026thinsp;=\u0026thinsp;+\u0026thinsp;0.106), total stone burden (r\u0026thinsp;=\u0026thinsp;+\u0026thinsp;0.101), and age (r\u0026thinsp;=\u0026thinsp;+\u0026thinsp;0.045). These results suggest that LP stones are more common in slightly older patients and in those with greater stone volume but are also related to a higher risk of incomplete clearance and RF, compared to the general population. In contrast, the presence of multiple stones (r = \u0026minus;\u0026thinsp;0.154), haematuria (r = \u0026minus;\u0026thinsp;0.138), pre-stenting (r = \u0026minus;\u0026thinsp;0.135), and use of a ureteral access sheath (UAS; r = \u0026minus;\u0026thinsp;0.128) were negatively associated with LP stones. This implies that LP stones tend to be solitary, less symptomatic, and can be often managed without UAS or prior stenting (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThere was no significant difference in stone density between the two groups, due to the fact that HU are mostly determined by stone composition and not stone location.\u003c/p\u003e\u003cp\u003ePostoperative complications did not show significant patterns linked with the presence of LP stones. Infectious complications like fever (r = -0.091) and sepsis (r = -0.072) showed only mild negative correlation with LP stones, and similar results were found for the presence of positive preoperative MSU (r = -0.054).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eMain results of correlation analysis, divided in negative and positive correlations (with relative r values).\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003ePositive correlations\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u003cp\u003eNegative correlations\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFeature\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCorrelation\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFeature\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eCorrelation\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRF\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.106\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eStone number\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.154\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTotal stone burden (mm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.101\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHaematuria\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.138\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eReintervention\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.047\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePresented\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.135\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge (years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.045\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eUAS used\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.128\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHolmLP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.044\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLaser time (minutes)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.076\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eReusable scope\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.040\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFever\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.091\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eKnown genetic disorders\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.038\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSepsis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.072\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePostoperative stent\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.034\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePositive MSU\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.054\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNormal kidney anatomy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e+\u0026thinsp;0.012\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMultiple fragments\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u0026minus;0.032\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eA hierarchical clustering analysis identified feature groups with tight interdependencies, notably between UAS use and sheath size (VIF\u0026thinsp;\u0026gt;\u0026thinsp;70), warranting the removal of one to reduce multicollinearity. Pairwise comparisons confirmed that LP stones were more frequently associated with single stones, lower haematuria rates, and reduced use of UAS and larger sheaths.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eA detailed overview of the correlation and ML analysis in available in the supplementary materials section.\u003c/p\u003e\u003cp\u003e\u003cem\u003eMachine Learning Model Performance\u003c/em\u003e\u003c/p\u003e\u003cp\u003eFifteen machine learning classifiers were trained using preoperative variables to predict the presence of LP stones. Ensemble-based models significantly outperformed individual classifiers.\u003c/p\u003e\u003cp\u003eRandom Forest emerged as the top-performing model with a validation accuracy of 80.95%, precision of 80.63%, recall of 75.00%, and F1-score of 76.67%, reflecting excellent balance between sensitivity and specificity. Gradient Boosting demonstrated similarly robust generalization with 76.19% validation accuracy and balanced metric profiles (training accuracy: 97.06%). Extra Trees Classifier reached 73.81% validation accuracy but showed signs of overfitting due to perfect training performance.\u003c/p\u003e\u003cp\u003eOther strong models included CatBoost and Bagging Classifier, both achieving 71.43% validation accuracy. In contrast, Quadratic Discriminant Analysis (QDA) and Na\u0026iuml;ve Bayes underperformed (validation accuracy\u0026thinsp;~\u0026thinsp;40\u0026ndash;50%), while Support Vector Machine variants displayed modest accuracy (~\u0026thinsp;66.67%) but low F1-scores, indicating poor class boundary definition.\u003c/p\u003e\u003cp\u003e\u003cem\u003eFeature Importance and Explainability\u003c/em\u003e\u003c/p\u003e\u003cp\u003eExplainable AI techniques, including SHAP (SHapley Additive exPlanations) and decision tree visualizations, provided interpretability of model predictions.\u003c/p\u003e\u003cp\u003eTop predictors of LP stone presence included: Stone number (single vs. multiple); Total operative time; Maximum stone diameter; Total stone burden; Age; Residual Fragments; UAS usage; and Presentation status (pre-stenting).\u003c/p\u003e\u003cp\u003eSHAP summary plots indicated that multiple stones most consistently increased the model\u0026rsquo;s likelihood of predicting LP stones, and this stone location was linked to shorter laser times and higher risk of RF. Conversely, features such as female sex, avoiding UAS use, and shorter operative time contributed negatively, reducing LP stone probability in specific predictions.\u003c/p\u003e\u003cp\u003eDecision tree visualizations confirmed that stone number was the initial node in model splitting, with subsequent branches involving RF, stone diameter, and fragmentation patterns. This hierarchy supported the clinical assumption that solitary stones and minimal instrumentation are more typical in LP stone cases.\u003c/p\u003e\u003cp\u003ePrediction-level explanations (via SHAP waterfall and force plots) confirmed that individual outcomes were strongly driven by procedural details and stone configuration, rather than isolated demographic or biochemical features.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eSummary of Predictive Accuracy\u003c/em\u003e\u003c/p\u003e\u003cp\u003eOverall, ensemble models, especially Random Forest and Gradient Boosting demonstrated superior performance in identifying LP stones, with validation accuracies exceeding 75%. While individual predictors showed weak linear correlation with LP stone presence, their collective interaction in nonlinear models yielded clinically meaningful predictive power. These results highlight the value of ML in supporting diagnostic decision-making for paediatric urolithiasis.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe ML model trained well on the retrospective database, providing good accuracy in prediction of LP related features and outcomes. To our knowledge, this is currently the largest ML analysis on paediatric patients undergoing fURSL, and the only one to investigate the role of LP stones in probability of stone clearance and risk of complications.\u003c/p\u003e\u003cp\u003eAs suspected, we found an increased risk of incomplete clearance in this population, alongside the need for multiple intervention in light of both the risk of residual fragments and recurrence of stones. Interestingly, infectious complications appear to be slightly less frequent in children with LP stones compared to other stone locations, and this is reflected by the lower rates of positive urine cultures and patients needing extended preoperative antibiotics. Postoperative haematuria was also less frequently encountered in this population group, despite the well-known challenges that a stiff infundibular angle can pose while treating LP stones [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. In our opinion, these findings might somehow be related, given the increased risk of contact-bleeding and mucosal abrasion during laser treatment when dealing with inflamed tissues, characteristics of patients with recurrent urinary tract infections or infected stones.\u003c/p\u003e\u003cp\u003eBased on our analysis, preoperative presentations and stone characteristics can differ in children affected by LP and non-LP stones. Older children and adolescents tend to have larger stones located in the LP, but they often present with single stones that are not obstructive and therefore require preoperative stenting less frequently. These findings may relate with renal anatomy, with gravity and infundibular angle playing a role in the larger growth of urolithiasis in LP stones before they get symptomatic and prompt interventions. Similarly, we found a moderate correlation between LP stones and known genetic disorders, with increased risk of urine stasis and stone formation. This link appears to be stronger that the one with anatomical variants of the kidney.\u003c/p\u003e\u003cp\u003ePrevious studies, not performed with the innovative aid of ML analysis, found an increased risk for incomplete clearance in children with LP stones. These results are in line with our analysis, with same risk of operative and postoperative complications reported in the two subgroups [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. As also described by Sen and colleagues, LP stones can be difficult to completely remove, but RF in these cases can frequently be asymptomatic and more rarely require reintervention [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Nevertheless, thanks to technological advancements and miniaturisation of fURSL instruments, ureteroscopy can now considered a first line treatment in children with LP stones, and a two-step procedure can still be offered as complication rates and impact on children quality of life appears to be improved compared to PCNL [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Previous systematic review looking at the role of fURSL for paediatric renal stones recommends that these procedures are done by experienced surgeons [17].\u003c/p\u003e\u003cp\u003eOur study represents one of the first ML-based analysis on a paediatric population undergoing fURSL for urolithiasis, and the first to describe findings on a large cohort with LP stones. Despite the retrospective nature, that comes with possible biases of missing or incomplete data, the analysis and ML training resulted in good performance and identification of predictive factors and outcomes. This multicentric database collected a large number of paediatric cases and reflects the variety of real-word practice. We believe that this represents a strength of the study, and with further external validation and prospective analysis a predictive model might be developed for this fragile subgroup with increased risk of incomplete stone clearance. Perhaps future studies should also look at the cost of fURSL procedures and adaption of ML techniques in outcome prediction for these procedures [18,19].\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eChildren with LP stones can be safely and effectively treated with fURSL, but a higher risk of incomplete stone clearance needs to be acknowledged. Obstructive stones are less frequent in patients with LP stones, and there is a reduced need for preoperative stent insertion. Complications rates appear to be similar to the general paediatric population. ML models demonstrated strong predictive performance and could support preoperative risk stratification, but further external validation and implementation with prospective data are warranted to refine predictive tools for clinical use.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConsent to Participate:\u0026nbsp;\u003c/strong\u003eall patients included in the study signed an informed consent for data collection and research analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics Approval:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003egiven the anonymous and retrospective nature of the study, no ethical approval was needed according to our Internal Review Board (IRB).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDeclaration: no Funding.\u003c/p\u003e\n\u003cp\u003eHuman Ethics and Consent to Participate declarations: not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCN: writing, editing, project conceptualisationVG, SKKY, EJL: data collection, project conceptualisation, editingSG, HT, RM, PJ, KK: data analysis, results, figuresAG, FA, FP, YT, AS, BS, HYL, NN: project conceptualisation, editingBKS: project conception, editing, supervisionAll authors reviewed the manuscript\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDeclarationsConsent to Participate: all patients included in the study signed an informed consent for data collection and research analysis.Ethics Approval: given the anonymous and retrospective nature of the study, no ethical approval was needed according to our Internal Review Board (IRB).Funding Declaration: no Funding.Human Ethics and Consent to Participate declarations: not applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eJobs K, Rakowska M, Paturej A (2018) Urolithiasis in the pediatric population - current opinion on epidemiology, patophysiology, diagnostic evaluation and treatment. Dev Period Med 22:201\u0026ndash;208. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.34763/devperiodmed.20182202.201208\u003c/span\u003e\u003cspan address=\"10.34763/devperiodmed.20182202.201208\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSkolarikos A, Neisius A, Petř\u0026iacute;k A et al (2022) EAU Guidelines on Urolithiasis. In: EAU guidelines. Edn. presented at the EAU Annual congress Amsterdam. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://uroweb.org/guidelines/urolithiasis/chapter/introduction\u003c/span\u003e\u003cspan address=\"https://uroweb.org/guidelines/urolithiasis/chapter/introduction\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTsai S-H, Chung H-J, Tseng P-T et al (2020) Comparison of the efficacy and safety of shockwave lithotripsy, retrograde intrarenal surgery, percutaneous nephrolithotomy, and minimally invasive percutaneous nephrolithotomy for lower-pole renal stones. Medicine 99:e19403. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1097/MD.0000000000019403\u003c/span\u003e\u003cspan address=\"10.1097/MD.0000000000019403\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFrascheri MF, Contreras P, Enikeev D et al (2025) Small tools, big impact and redefining endourology: the paradigm shift of ureteroscopy for large stones and percutaneous nephrolithotomy for small stones \u0026ndash; a literature review. Curr Opin Urol 35:347\u0026ndash;353. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1097/MOU.0000000000001283\u003c/span\u003e\u003cspan address=\"10.1097/MOU.0000000000001283\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKandemir A, Guven S, Balasar M et al (2017) A prospective randomized comparison of micropercutaneous nephrolithotomy (Microperc) and retrograde intrarenal surgery (RIRS) for the management of lower pole kidney stones. World J Urol 35:1771\u0026ndash;1776. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00345-017-2058-9\u003c/span\u003e\u003cspan address=\"10.1007/s00345-017-2058-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlma E, Altunkol A, Anıl H et al (2025) Comparision of scoring systems in the success of retrograde intrarenal surgery and factors affecting the presence of residual fragments in pediatric patients. J Pediatr Urol. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jpurol.2025.06.011\u003c/span\u003e\u003cspan address=\"10.1016/j.jpurol.2025.06.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eArikan Y (2024) In the Treatment of Lower Pole Kidney Stones Between 1\u0026ndash;2 cm in Children, Which is the Best Approach? Retrograde Intrarenal Surgery or Mini Percutaneous Nephrolithotomy. SiSli Etfal Hastanesi Tip Bulteni /. Med Bull Sisli Hosp 332\u0026ndash;338. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.14744/semb.2024.49225\u003c/span\u003e\u003cspan address=\"10.14744/semb.2024.49225\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim HJ, Daignault-Newton S, DiBianco JM et al (2023) Real-world Practice Stone-free Rates After Ureteroscopy: Variation and Outcomes in a Surgical Collaborative. Eur Urol Focus 9:773\u0026ndash;780. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.euf.2023.03.010\u003c/span\u003e\u003cspan address=\"10.1016/j.euf.2023.03.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMille E, El-Khoury E, Haddad M et al (2023) Comparison of single-use flexible ureteroscopes with a reusable ureteroscope for the management of paediatric urolithiasis. J Pediatr Urol 19:248e1\u0026ndash;248e6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jpurol.2023.01.009\u003c/span\u003e\u003cspan address=\"10.1016/j.jpurol.2023.01.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRosi\u0026egrave;re NI, Ruiz J, L\u0026oacute;pez Imizcoz F et al (2025) Predictive variables for successful retrograde intrarenal surgery outcomes in pediatric patients: Development of nomograms. J Pediatr Urol 21:775\u0026ndash;776. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jpurol.2025.01.012\u003c/span\u003e\u003cspan address=\"10.1016/j.jpurol.2025.01.012\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNedbal C, Adithya S, Gite S et al (2024) A Machine Learning Predictive Model for Ureteroscopy Lasertripsy Outcomes in a Pediatric Population\u0026mdash;Results from a Large Endourology Tertiary Center. J Endourol. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1089/end.2024.0120\u003c/span\u003e\u003cspan address=\"10.1089/end.2024.0120\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDong J, Feng T, Thapa-Chhetry B et al (2021) Machine learning model for early prediction of acute kidney injury (AKI) in pediatric critical care. Crit Care 25:288. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13054-021-03724-0\u003c/span\u003e\u003cspan address=\"10.1186/s13054-021-03724-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFernandez N, Ellison JS, Wang Z et al (2024) Surgeon, and Institution Characteristics Associated Surgical Preferences in the Pediatric KIDney Stone Care Improvement Network. Urology 187:64\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.urology.2024.02.040\u003c/span\u003e\u003cspan address=\"10.1016/j.urology.2024.02.040\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eUnal U, Deniz ME, Ortoglu F et al (2022) What is the best approach for pediatric kidney stones of moderate-sized between shock wave lithotripsy, ultramini percutaneous nephrolithotomy and retrograde intrarenal surgery? Pediatr Surg Int 38:1643\u0026ndash;1648. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00383-022-05203-1\u003c/span\u003e\u003cspan address=\"10.1007/s00383-022-05203-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSen H, Baydilli N, Ozturk M et al (2024) Factors effecting the success of retrograde intrarenal surgery in pediatric patients with renal stones: The experience of two tertiary centres with 368 renal units. J Pediatr Urol 20. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jpurol.2024.01.004\u003c/span\u003e\u003cspan address=\"10.1016/j.jpurol.2024.01.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. :403.e1-403.e9\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMosquera L, Pietropaolo A, Madarriaga YQ et al (2021) Is Flexible Ureteroscopy and Laser Lithotripsy the New Gold Standard for Pediatric Lower Pole Stones? Outcomes from Two Large European Tertiary Pediatric Endourology Centers. J Endourol 35:1479\u0026ndash;1482. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1089/end.2020.1123\u003c/span\u003e\u003cspan address=\"10.1089/end.2020.1123\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\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":"world-journal-of-urology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"wjur","sideBox":"Learn more about [World Journal of Urology](https://link.springer.com/journal/345)","snPcode":"345","submissionUrl":"https://submission.nature.com/new-submission/345/3","title":"World Journal of Urology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Paediatric urology, Urolithiasis, Lower pole stones, Machine Learning, Predictive Tools, Stone Free Rates.","lastPublishedDoi":"10.21203/rs.3.rs-7225953/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7225953/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eIntroduction\u003c/h2\u003e\u003cp\u003eThe rising incidence of kidney stone disease in children presents growing clinical challenges, particularly in managing lower pole (LP) calculi, which are anatomically difficult to treat. Flexible ureteroscopy with laser lithotripsy (fURSL) has emerged as a preferred minimally invasive treatment. However, surgical outcomes remain variable, especially in the paediatric LP stone cohort. This study aimed to apply machine learning (ML) techniques to predict surgical outcomes based on preoperative characteristics and identify key predictors of incomplete stone clearance.\u003c/p\u003e\u003ch2\u003eMaterials and Methods\u003c/h2\u003e\u003cp\u003eA retrospective analysis was conducted on paediatric patients (\u0026lt;\u0026thinsp;18 years) who underwent fURSL between January 2017 and December 2021 across eight tertiary centres. From a multicentre database of 280 patients, 91 with isolated LP stones were selected. Preoperative, intraoperative, and postoperative variables were analysed. Fifteen ML models\u0026mdash;including ensemble algorithms and a multitask neural network\u0026mdash;were developed to predict LP stone presence and postoperative outcomes. Model performance was evaluated using accuracy, precision, recall, F1-score, and SHAP (SHapley Additive exPlanations) values for interpretability.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eLP stones were present in 32.5% of cases and were associated with older age, solitary stones, and higher stone burden. Random Forest outperformed all other models (validation accuracy: 80.95%; F1-score: 76.67%), followed by Gradient Boosting. SHAP analysis identified stone number, total stone burden, age, and operative time as top predictors. LP stones were associated with a higher rate of residual fragments (RF) and lower need for preoperative stenting or ureteral access sheath use. Infectious and bleeding complications were less frequent in the LP group.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003efURSL is safe and effective in children with LP stones, though incomplete stone clearance remains a challenge. ML models demonstrated strong predictive performance and could support preoperative risk stratification. Further external validation and prospective studies are warranted to refine predictive tools for clinical use.\u003c/p\u003e","manuscriptTitle":"Efficacy and safety of ureteroscopy in children with lower pole renal stones – a machine learning predictive model from the EAU section of Endourology","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-11 10:03:21","doi":"10.21203/rs.3.rs-7225953/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-20T07:43:27+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-23T10:32:15+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-13T01:16:34+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"67014568694012205152809666469666108296","date":"2025-08-09T01:40:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"80801740863840702948764025090784368737","date":"2025-08-08T00:04:10+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"207147452536145055163894822371659499864","date":"2025-08-06T07:47:51+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-05T21:24:38+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-29T23:44:23+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-29T17:53:09+00:00","index":"","fulltext":""},{"type":"submitted","content":"World Journal of Urology","date":"2025-07-27T11:28:16+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"world-journal-of-urology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"wjur","sideBox":"Learn more about [World Journal of Urology](https://link.springer.com/journal/345)","snPcode":"345","submissionUrl":"https://submission.nature.com/new-submission/345/3","title":"World Journal of Urology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"3d132182-1857-4ddc-a311-3bb06557f4e9","owner":[],"postedDate":"August 11th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-11-24T16:07:45+00:00","versionOfRecord":{"articleIdentity":"rs-7225953","link":"https://doi.org/10.1007/s00345-025-06095-1","journal":{"identity":"world-journal-of-urology","isVorOnly":false,"title":"World Journal of Urology"},"publishedOn":"2025-11-20 15:58:55","publishedOnDateReadable":"November 20th, 2025"},"versionCreatedAt":"2025-08-11 10:03:21","video":"","vorDoi":"10.1007/s00345-025-06095-1","vorDoiUrl":"https://doi.org/10.1007/s00345-025-06095-1","workflowStages":[]},"version":"v1","identity":"rs-7225953","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7225953","identity":"rs-7225953","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}
Text is read by the "Ask this paper" AI Q&A widget below. Extraction quality varies by source — PMC NXML preserves structure cleanly, OA-HTML may include some navigation residue, and OA-PDF can have broken hyphenation. The publisher copy (via DOI) is the canonical version.
My notes (saved in your browser only)
Ask this paper
Answers must be backed by verbatim quotes from this paper's full text. Hallucinated quotes are dropped automatically; if no verbatim passage answers the question, we say so. How this works
Citation neighborhood (no data yet)
We don't have any in-corpus citations linked to this paper yet. This is a recent paper (2025) — citers typically take a year or two to land, and the OpenAlex reference graph may still be filling in.
Source provenance
- europepmc
- last seen: 2026-05-20T01:45:00.602351+00:00
- unpaywall
- last seen: 2026-05-29T02:00:03.542394+00:00
License: CC-BY-4.0