Risk factors and predictive models for perioperative acute kidney injury in children: a narrative review

Introduction

Acute kidney injury (AKI) is a rapid decline in kidney function, and can occur in neonates, children, and adults. In pediatric patients, AKI is defined by an increase of serum creatinine (SCr) by 0.3 mg/dL or more within 48 h, an increase of SCr to 1.5-times or more over the baseline level within the prior 7 days, or a decrease in urine output to below 0.5 mL/kg/h for 6 h (1). The neonatal modified KDIGO (nKDIGO) definition is commonly used in neonates, and it uses specific thresholds for SCr based on postnatal age (2). AKI is associated with increased morbidity and mortality, particularly in vulnerable populations, such as children during the perioperative period (3). In particular, the mortality rate can exceed 20% in neonates undergoing major surgery and there is a 2.5-fold increased risk of long-term mortality for affected children compared to non-AKI cohorts (4,5).

Epidemiological studies indicate that pediatric populations have a greater risk for AKI during the perioperative period, and the reported incidence ranges from 5% to 30% depending on the type of surgery and underlying health conditions. Dehydration, exposure to nephrotoxic agents, and hemodynamic instability during surgery increase this risk (6,7). Recent studies have highlighted that neonates and infants are particularly susceptible to perioperative AKI, and that this complication occurs more frequently in those undergoing cardiac surgery and other complex procedures (8). Neonates, particularly those admitted to intensive care, are highly susceptible to perioperative AKI due to renal immaturity, the smaller number of nephrons in low-birth-weight and preterm infants, and associated comorbidities. Furthermore, survivors of pediatric AKI face a substantial risk of developing chronic kidney disease (CKD); approximately 30% of these patients develop CKD within 3 to 5 years, and this increases the lifetime risk of cardiovascular disease (9).

The purpose of this review is to synthesize current knowledge about the epidemiology, risk factors, and outcomes associated with perioperative AKI in children. We aim to highlight the importance of early recognition and early implementation of interventions that mitigate the impact of AKI on children. We present this article in accordance with the Narrative Review reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0238/rc).

Methods

A comprehensive literature search was conducted to identify relevant studies on perioperative AKI in pediatric populations, with a specific focus on risk factors and prediction models. The search strategy aimed to be comprehensive within the scope of a narrative review, capturing seminal and recent publications to provide a current and scholarly overview. The detailed literature search strategy is summarized in Table 1.

Consideration of AKI definitions and literature synthesis

The use of different definitions for AKI is a persistent challenge, particularly for neonatal populations. To analyze a broad range of evidence, our review included studies that employed any established definition. During data extraction and synthesis, the specific AKI definition used in each study was documented. The potential impact of using different definitions of AKI on the reported epidemiology and performance of prediction models was considered and discussed as a key challenge in this review.

Scope and inclusivity of the search

To avoid omitting foundational publications, the search strategy did not impose a start date and considered all relevant publications (original research, reviews, meta-analyses) that were published up to March 2024 (Table 1). This allowed the inclusion of seminal older studies and cutting-edge research from 2023 to 2024, fulfilling the goal of providing a current yet historically contextualized overview.

Epidemiology of perioperative AKI in children

Incidence of AKI in children

AKI is a significant clinical condition in pediatric populations, particularly in perioperative settings. The reported incidence of perioperative AKI in children typically ranges from 5% to 30%, depending on the surgical procedure and the specific population. For instance, AKI is particularly prevalent in neonates and infants undergoing cardiac surgery, with incidence rates reported as high as 40% (6). The factors responsible for this high prevalence include the physiological immaturity of the neonatal renal system and the hemodynamic changes associated with surgical procedures. The use of nephrotoxic agents, such as certain anesthetics and antibiotics, also increases the risk of perioperative AKI (5). The early identification, vigilant monitoring, and prompt treatment of AKI are critical, because these patients can experience increased morbidity and prolonged hospital stays.

Risk of AKI in different pediatric age groups

The risk of developing perioperative AKI in pediatric patients varies among different age groups. Infants, particularly those less than 1-year-old, have a higher risk than older children. This increased risk in infants can be attributed to their immature kidneys, higher fluid turnover, increased exposure to nephrotoxic medications during critical periods of development, and other factors (10). Although older children and adolescents have a lower incidence of AKI, they are still vulnerable, especially those who receive complex surgeries or who have certain underlying health conditions. For example, a recent study found that the risk factors for perioperative AKI in older children included pre-existing renal dysfunction and dehydration (11). An improved understanding of the epidemiology and pathophysiology of these age-related differences is crucial for improving preventive strategies and perioperative care.

Long-term prognosis of children with AKI

There is increasing recognition of the long-term consequences of AKI in children, and research indicates that children who experience perioperative AKI may have an increased risk of CKD later in life. More specifically, a systematic review concluded that approximately 30% of children with AKI may experience persistent renal impairment and require long-term follow-up (12). Additional factors, such as the severity of AKI, the presence of comorbidities, and patient age at the time of AKI onset, can significantly affect long-term outcome. For instance, children who develop severe perioperative AKI during infancy have a greater risk of adverse kidney-related outcomes than children who experience mild AKI later in childhood (13). Therefore, proactive monitoring and management strategies are essential to improve the long-term renal health of children who experience AKI.

Risk factors for perioperative AKI in children

Age and gender

Age and gender significantly influence the risk of AKI during the period of pediatric perioperative care. In particular, younger children, particularly neonates and infants, have a greater risk for developing AKI due to their immature renal function and limited physiological reserves. For instance, a 2024 study highlighted that perioperative AKI was common in neonates undergoing cardiac surgery, and that young age was a critical determinant of poor outcomes (6). Furthermore, other research suggested that male children may have a higher susceptibility to AKI than female children, potentially due to sex-related differences in renal physiology and hormonal effects (5). A better understanding of these demographic factors may improve risk stratification and the selection of targeted preventive strategies in these pediatric surgical patients.

Underlying medical conditions

The presence of underlying medical conditions can also significantly increase the risk of AKI in children undergoing surgery, and surgery for congenital heart disease (CHD) greatly increases the risk of perioperative AKI. Children receiving surgery for CHD often present with compromised renal perfusion and an overall impairment of renal function, making them more vulnerable during the perioperative period (7). Pre-existing renal diseases, such as congenital anomalies of the kidney and urinary tract (CAKUT), can also increase the risk of pediatric AKI because these individuals have reduced renal reserve and altered hemodynamics during surgery (8). The preoperative identification of these underlying conditions can aid in developing more tailored anesthesia and surgical plans to mitigate the risk of AKI.

Type of surgery and anesthesia techniques

The type of surgical procedure and anesthesia technique are critical determinants of pediatric AKI. Certain high-risk surgeries, such as cardiac operations that employ cardiopulmonary bypass, are closely associated with perioperative AKI due to the presence of hemodynamic instability and inflammatory responses (14). The choice of anesthesia can also influence renal outcomes. For example, dexmedetomidine can reduce the incidence of perioperative AKI in pediatric patients receiving cardiac surgery, suggesting that it is necessary for anesthesiologists and clinicians to carefully consider anesthetic agents to prevent renal injury (15). Thus, careful consideration of surgical techniques and anesthetic approaches are crucial for minimizing the risk of perioperative AKI in children.

Preoperative and intraoperative use of medications

The use of medications in the perioperative setting, particularly nonsteroidal anti-inflammatory drugs (NSAIDs) and certain antibiotics, can significantly increase the risk of pediatric AKI. Although NSAIDs are effective for pain management, their inhibition of prostaglandin synthesis can impair renal function and inhibit renal perfusion, especially in pediatric patients who are already vulnerable (16). Additionally, nephrotoxic antibiotics, such as aminoglycosides, increase the risk for AKI, particularly when used in conjunction with other nephrotoxic agents or in patients with pre-existing renal impairment (17). Careful evaluation of medications and consideration of appropriate alternatives or dosing adjustments can help to reduce the risk of perioperative AKI in children.

In summary, a more complete understanding of the many factors associated with perioperative AKI in children is crucial for improving outcomes and guiding clinical practices.

Development of risk prediction models

Overview of existing models

Risk prediction models are essential tools for clinical practitioners because they can improve estimates of the likelihood of specific outcomes based on patient characteristics. Although there are several validated risk stratification tools for adults, their applicability to pediatric populations remains limited. For instance, the Cleveland Clinic Scoring System predicts AKI in adults undergoing cardiac surgery by integration of variables such as preoperative estimated glomerular filtration rate (eGFR) and type of surgery (18), and the Society of Thoracic Surgeons Operative Risk Calculator provides comprehensive profiles for the risk of AKI complications in adult patients (19). However, these models have not been validated in children, who have significantly different physiology and risk factors. Although there are very few risk prediction models that focus on perioperative AKI in children, researchers have developed risk prediction models for many other specific fields in medicine, each tailored to assess the risk associated with a particular condition. For instance, an overview of systematic reviews examined models for predicting AKI in adults, and highlighted the importance of incorporating multiple clinical variables to improve predictive accuracy (20). Other models have focused on other specific conditions, such as colorectal liver metastases, and evaluated the prognostic value of regression-based and machine learning (ML) approaches (21). Additionally, a 2024 systematic review concluded that the integration of ML techniques improved the predictive power of traditional statistical models, and provided more nuanced assessments of risk in complex patient populations (22). Although existing models vary in their complexity and application, they all have the common goal of improving clinical decision-making and patient outcomes.

Application of statistical methods and ML

The development of risk prediction models has evolved significantly with the advent of ML techniques, and these new techniques are a significant complement to traditional statistical methods. Statistical methods, such as logistic regression, have long been used to analyze relationships between variables and outcomes, and they provide a solid foundation for model development (23). However, ML approaches have the capability to analyze large datasets and uncover complex patterns and relationships that may not be apparent from conventional statistical analysis (24). For example, ensemble models, which use multiple ML algorithms, are particularly effective in predicting risks associated with different surgical procedures, thereby enhancing the precision of risk assessment (25). Furthermore, the integration of statistical and ML methods can facilitate the selection of biomarkers and improve the accuracy of cardiovascular risk predictions (26). As these methodologies continue to evolve, their application will likely provide more robust and reliable risk prediction models that can be used in clinical settings.

Model validation, calibration, and clinical applicability

The validation of a risk prediction model is a critical step in ensuring its clinical applicability and reliability. Validation requires assessment of a model’s performance in diverse populations and clinical settings to confirm its predictive accuracy and generalizability. For instance, external validation studies have evaluated models designed to predict severe hypoglycemia, thus emphasizing the need for rigorous testing before clinical implementation (27). Moreover, it is important to assess the calibration of a risk assessment model to ensure that the predicted probabilities have a close match with the observed outcomes (28). Clinical applicability—a model’s usability in real-world scenarios—must also be considered to ensure the ease of model interpretation and integration into existing clinical workflows. As the healthcare landscape continues to evolve, the focus on developing validated and clinically applicable risk prediction models will remain paramount for improving patient care and outcomes.

Clinical application of risk prediction models

Implementing risk assessment in clinical settings

The implementation of reliable risk assessment tools in clinical settings is crucial for improving patient outcomes and optimizing healthcare delivery. Reliable risk assessment requires the integration of validated prediction models into routine clinical practice, and this requires the training of healthcare professionals to use these tools and to then accurately interpret the results. For instance, there is evidence that the successful implementation of breast cancer risk assessment programs can lead to improved patient engagement and more informed decision-making regarding screening and preventive measures (29). Moreover, the incorporation of risk assessment tools into clinical settings must consider the specific context, including patient demographics and the local prevalence of the targeted disease. Before use of a model, it is essential to ensure that the model is clinically valid and practical (30). Additionally, the adaptation of models to accommodate electronic health records can streamline their application and improve their accessibility for clinicians (31). Overall, the successful implementation of risk assessment tools requires a multifaceted approach that includes education, contextual adaptation, and integration of technologies.

Impact of risk prediction models on clinical decision-making

Risk prediction models can significantly improve clinical decision-making because they provide evidence-based estimates of patient outcomes. More specifically, these models enable clinicians to stratify patients according to risk, so that more tailored interventions can be implemented to enhance patient safety and resource allocation. For example, a meta-analysis of diabetic foot ulcers showed that predictive models were effective in the identification of high-risk patients, and this facilitated more timely interventions and reduced the incidence of complications (32). Furthermore, clinical prediction models can improve outcomes for many conditions, including conditions related to maternal health, such as the risk of postpartum depression (33). However, the effectiveness of these models is contingent upon their accuracy and a clinician’s ability to interpret the results within the specific clinical context (34). Thus, although risk prediction models are valuable tools for clinical decision-making, they are even more beneficial when they are combined with clinical judgments and patient-specific considerations.

The recent advancements in development of pediatric AKI prediction models are characterized by an evolution from models that consider static variables to those that consider dynamic variables (35). For instance, a neural network model predicted AKI based on a 50% or more increase of SCr within 24 h of admission had an overall accuracy of 68.1% (95% confidence interval: 67.6%, 68.7%), its accuracy improved to 90 to 96% in children with baseline SCr of 0.6 mg/dL or more, and it had a strong negative predictive value (97.2%) for extreme elevation of SCr (36). This model utilizes 36 routinely available variables, including demographics, baseline SCr, platelet count, and vasoactive medication status. Another prospectively validated random forest model that incorporated 33 admission variables demonstrated high discrimination in multicenter validation (area under the curve of 0.929 for internal validation and 0.910 for external validation), and SHAP analysis identified eight key predictors of AKI, including baseline SCr, platelet count, and oxygenation index (37). Notably, most models do not incorporate dynamic physiological parameters. A systematic review of 161 externally validated models showed that only 12.4% of them considered dynamic variables, despite significant performance variations among clinical subgroups (38). Future developments should prioritize adaptive algorithms that can process real-time physiological data to enhance generalizability.

Development and implementation of preventive measures

The development and implementation of preventive measures based on risk prediction models are essential for improving patient care and reducing adverse outcomes. When healthcare providers can identify patients who have a high risk for a specific condition, they can then initiate personalized preventive strategies. For instance, in the field of cardiovascular health, risk models can guide the implementation of lifestyle and pharmacological interventions that reduce the risk of heart failure in patients with type 2 diabetes (39). Additional preventive measures can be further enhanced by utilizing patient education and engagement strategies that empower the patient to take a more active role in health management (40). However, the success of these preventive measures depends on continuous evaluation and adaptation of the model based on real-world outcomes, and this necessitates ongoing research and clinical feedback (41). Ultimately, the integration of risk prediction models to guide the interventions used to treat and prevent diseases has significant potential to improve health outcomes and reduce healthcare costs.

Directions for future research

Identification of novel biomarkers

The identification and validation of novel biomarkers for perioperative AKI in children is a critical area for future research. Current biomarkers, such as SCr and urine output, are often inadequate for early detection and precise risk stratification of children. Instead, recent studies have highlighted the potential of various novel biomarkers, including neutrophil gelatinase-associated lipocalin (NGAL), interleukin-18 (IL-18), and kidney injury molecule-1 (KIM-1), for the early detection of AKI and predicting patient outcomes (42). Additionally, the integration of novel biomarkers derived from genomic and proteomic studies may enhance our understanding of the pathophysiology of pediatric AKI and provide insights into more personalized treatment strategies (43). As this research progresses, it will also be essential to conduct large-scale validation studies to confirm the clinical utility of these different biomarkers in diverse pediatric populations and settings, so that they can be reliably adopted in clinical practice to improve patient outcomes (44).

In addition to these biomarkers, models that consider urinary liver fatty acid binding protein (L-FABP) and the combination of tissue inhibitor of metalloproteinase-2 and insulin-like growth factor-binding protein 7 (TIMP-2×IGFBP7) are in advanced stages of validation for early detection of postoperative pediatric AKI. More specifically, recent research demonstrated that urinary L-FABP had high diagnostic accuracy for pediatric AKI following cardiac surgery with cardiopulmonary bypass, particularly in identifying subclinical injury at 24 to 48 h before serum SCr elevation (45). Similarly, TIMP-2×IGFBP7, which reflects cell-cycle arrest, is valuable for predicting severe AKI in neonates and infants undergoing cardiac procedures, and its performance was validated in multicenter cohorts (46).

Multicenter research and data sharing

Multicenter research initiatives are vital for advancing our understanding of perioperative AKI in children, because they collect diverse data across various demographics and clinical settings and can be used for data sharing. Collaborative studies of pediatric AKI can help identify variations in the incidence, risk factors, and outcomes, leading to more robust conclusions (47). Furthermore, the sharing of data among research institutions can increase the statistical power of a model and facilitate the development of comprehensive databases that track patient outcomes over time. These approaches thus improve our understanding of the epidemiology of perioperative AKI in children, and also foster the identification of best practices and most effective interventions (48). By leveraging advanced data-sharing technologies and frameworks, researchers can ensure the efficient use of clinical data, and also promote a culture of collaboration and innovation in pediatric nephrology (49).

Personalized medicine for management of pediatric AKI

The future management of pediatric AKI lies in the implementation of personalized medicine approaches that consider genetic predispositions, comorbidities, and specific AKI phenotypes. Recent findings suggest that different clinical phenotypes of AKI in critically ill children are associated with distinct outcomes, highlighting the need for more tailored management strategies (50). The use of personalized medicine often enables better predictions of patients who have a higher risk for progression to CKD, so that more appropriate targeted interventions can be used (51). Additionally, incorporating pharmacogenomics into the management of pediatric AKI may improve drug efficacy and decrease adverse effects. As research continues to evolve in this direction, it is imperative to integrate findings from multicenter studies to refine and validate personalized approaches in clinical practice, because this will likely improve the outcomes of children with AKI.

The development of risk prediction models is a critical advancement in this field, because these models are valuable tools for the early identification of children who have an elevated risk for AKI. These models can integrate clinical, demographic, and laboratory parameters, and enable healthcare providers to implement targeted preventive strategies that improve perioperative care. However, it is also essential to recognize that these models have certain limitations. These include the use of different definitions of AKI, variations among study populations, and the evolving nature of clinical practice, each of which can impact model applicability in different settings.

Limitations regarding the generalizability of evidence

We acknowledge that several key sections of this review, particularly those discussing prediction models, statistical methods, and clinical implementation, have relied on studies of adults. This is simply because most validated perioperative AKI prediction models have been developed and tested in adult populations; studies of AKI in pediatric patients remain limited in number, sample size, and external validation. Consequently, our ability to provide a comprehensive, pediatric-focused synthesis was constrained by the limited number of publications. We have made efforts to highlight pediatric evidence when present, but caution is needed when extrapolating findings from adult models to children, especially neonates and infants. Future research should prioritize multicenter pediatric studies to fill this critical evidence gap.

Conclusions

A look toward the future management of pediatric patients with AKI suggests a need for additional studies that refine existing risk prediction models and identify novel biomarkers, and for development of therapeutic interventions that improve prediction of perioperative AKI and patient outcomes. Balancing the insights provided by research that has different perspectives will be key to establishing a unified approach for prevention and management of AKI in children, and for ensuring these children receive the best possible care during the perioperative period.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the Narrative Review reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0238/rc

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0238/prf

Funding: This work was supported by Guangzhou Science and Technology Project (No. 2025A03J3709) and Guangdong Medical Association Research Fund (No. GDMA2024123106).

Conflicts of Interest: Both authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0238/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Rivetti G, Gizzone P, Petrone D, et al. Acute Kidney Injury in Children: A Focus for the General Pediatrician. Children (Basel) 2024;11:1004. [Crossref] [PubMed]
2. Jetton JG, Askenazi DJ. Update on acute kidney injury in the neonate. Curr Opin Pediatr 2012;24:191-6. [Crossref] [PubMed]
3. He L, Gao T, Tang Y, et al. A machine learning-based model for predicting the postoperative risk of acute kidney injury in neonates. Transl Pediatr 2025;14:3349-60. [Crossref] [PubMed]
4. Mahowald M, Khayat M. Kidney Disease: Acute Kidney Injury. FP Essent 2021;509:11-9.
5. Ojo B, Campbell CH. Perioperative acute kidney injury: impact and recent update. Curr Opin Anaesthesiol 2022;35:215-23. [Crossref] [PubMed]
6. Mohamed TH, Mpody C, Nafiu O. Perioperative Neonatal Acute Kidney Injury Is Common: Risk Factors for Poor Outcomes. Am J Perinatol 2024;41:e2818-23. [Crossref] [PubMed]
7. Hollander SA, Chung S, Reddy S, et al. Intraoperative and Postoperative Hemodynamic Predictors of Acute Kidney Injury in Pediatric Heart Transplant Recipients. J Pediatr Intensive Care 2024;13:37-45. [Crossref] [PubMed]
8. Sun M, Yang L, Zong Q, et al. Serum soluble triggering receptor levels expressed on myeloid cells2 identify early acute kidney injury in infants and young children after pediatric cardiopulmonary bypass. Front Pediatr 2023;11:1185151. [Crossref] [PubMed]
9. Darmon M, Truche AS, Abdel-Nabey M, et al. Early Recognition of Persistent Acute Kidney Injury. Semin Nephrol 2019;39:431-41. [Crossref] [PubMed]
10. Hui WF, Chan VPY, Cheung WL, et al. Risk factors for development of acute kidney injury and acute kidney disease in critically ill children. J Nephrol 2023;36:1425-34. [Crossref] [PubMed]
11. Bjornstad EC, Muronya W, Smith ZH, et al. Incidence and epidemiology of acute kidney injury in a pediatric Malawian trauma cohort: a prospective observational study. BMC Nephrol 2020;21:98. [Crossref] [PubMed]
12. Robinson CH, Jeyakumar N, Luo B, et al. Long-Term Kidney Outcomes Following Dialysis-Treated Childhood Acute Kidney Injury: A Population-Based Cohort Study. J Am Soc Nephrol 2021;32:2005-19. [Crossref] [PubMed]
13. Onal O, Chhabada S, Pu X, et al. Mild acute kidney injury after pediatric surgery is not-associated with long-term renal dysfunction: A retrospective cohort study. J Clin Anesth 2022;83:110985. [Crossref] [PubMed]
14. Yoneyama F, Okamura T, Takigiku K, et al. Novel Urinary Biomarkers for Acute Kidney Injury and Prediction of Clinical Outcomes After Pediatric Cardiac Surgery. Pediatr Cardiol 2020;41:695-702. [Crossref] [PubMed]
15. Kim EH, Lee JH, Kim HS, et al. Effects of intraoperative dexmedetomidine on the incidence of acute kidney injury in pediatric cardiac surgery patients: A randomized controlled trial. Paediatr Anaesth 2020;30:1132-8. [Crossref] [PubMed]
16. Shi S, Xiong C, Bie D, et al. Association between postoperative ibuprofen exposure and acute kidney injury after pediatric cardiac surgery. Ren Fail 2024;46:2318417. [Crossref] [PubMed]
17. Young AM, Strobel RJ, Rotar EP, et al. Perioperative acetaminophen is associated with reduced acute kidney injury after cardiac surgery. J Thorac Cardiovasc Surg 2024;167:1372-80. [Crossref] [PubMed]
18. Vives M, Candela A, Monedero P, et al. Improving the performance of the Cleveland Clinic Score for predicting acute kidney injury after cardiac surgery: a prospective multicenter cohort study. Minerva Anestesiol 2024;90:245-53. [Crossref] [PubMed]
19. Shahian DM, Jacobs JP, Badhwar V, et al. The Society of Thoracic Surgeons 2018 Adult Cardiac Surgery Risk Models: Part 1-Background, Design Considerations, and Model Development. Ann Thorac Surg 2018;105:1411-8. [Crossref] [PubMed]
20. Van Acker P, Van Biesen W, Nagler EV, et al. Risk prediction models for acute kidney injury in adults: An overview of systematic reviews. PLoS One 2021;16:e0248899. [Crossref] [PubMed]
21. Kokkinakis S, Ziogas IA, Llaque Salazar JD, et al. Clinical Prediction Models for Prognosis of Colorectal Liver Metastases: A Comprehensive Review of Regression-Based and Machine Learning Models. Cancers (Basel) 2024;16:1645. [Crossref] [PubMed]
22. Zhang M, Zheng Y, Maidaiti X, et al. Integrating Machine Learning into Statistical Methods in Disease Risk Prediction Modeling: A Systematic Review. Health Data Sci 2024;4:0165.
23. Staffa SJ, Zurakowski D. Statistical Development and Validation of Clinical Prediction Models. Anesthesiology 2021;135:396-405. [Crossref] [PubMed]
24. Tomašev N, Glorot X, Rae JW, et al. A clinically applicable approach to continuous prediction of future acute kidney injury. Nature 2019;572:116-9. [Crossref] [PubMed]
25. Grantham JP, Hii A, Shenfine J. Combined and intraoperative risk modelling for oesophagectomy: A systematic review. World J Gastrointest Surg 2023;15:1485-500. [Crossref] [PubMed]
26. Sperti M, Malavolta M, Staunovo Polacco F, et al. Cardiovascular risk prediction: from classical statistical methods to machine learning approaches. Minerva Cardiol Angiol 2022;70:102-22. [Crossref] [PubMed]
27. Yun JS, Han K, Choi SY, et al. External validation and clinical application of the predictive model for severe hypoglycemia. Front Endocrinol (Lausanne) 2022;13:1006470. [Crossref] [PubMed]
28. Trottini M, Campus G, Corridore D, et al. Assessing the Predictive Performance of Probabilistic Caries Risk Assessment Models: The Importance of Calibration. Caries Res 2020;54:258-65. [Crossref] [PubMed]
29. Silver E, Wenger N, Xie Z, et al. Implementing a Population-Based Breast Cancer Risk Assessment Program. Clin Breast Cancer 2019;19:246-253.e2. [Crossref] [PubMed]
30. Chaimowitz GA, Mamak M, Moulden HM, et al. Implementation of risk assessment tools in psychiatric services. J Healthc Risk Manag 2020;40:33-43. [Crossref] [PubMed]
31. Jawa V, Maamary J, Swanson M, et al. Implementing a Clinical Immunogenicity Strategy using Preclinical Risk Assessment Outputs. J Pharm Sci 2022;111:960-9. [Crossref] [PubMed]
32. Guo P, Tu Y, Liu R, et al. Performance of risk prediction models for diabetic foot ulcer: a meta-analysis. PeerJ 2024;12:e17770. [Crossref] [PubMed]
33. Qi W, Wang Y, Li C, et al. Predictive models for predicting the risk of maternal postpartum depression: A systematic review and evaluation. J Affect Disord 2023;333:107-20. [Crossref] [PubMed]
34. Goodacre S. Using clinical risk models to predict outcomes: what are we predicting and why? Emerg Med J 2023;40:728-30. [Crossref] [PubMed]
35. Collins GS, Reitsma JB, Altman DG, et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ 2015;350:g7594. [Crossref] [PubMed]
36. Dixon CG, Trujillo Rivera EA, Patel AK, et al. Development of a neural network model for early detection of creatinine change in critically Ill children. Front Pediatr 2025;13:1549836. [Crossref] [PubMed]
37. Hu J, Xu J, Li M, et al. Identification and validation of an explainable prediction model of acute kidney injury with prognostic implications in critically ill children: a prospective multicenter cohort study. EClinicalMedicine 2024;68:102409. [Crossref] [PubMed]
38. Rehman AU, Neyra JA, Chen J, et al. Machine learning models for acute kidney injury prediction and management: a scoping review of externally validated studies. Crit Rev Clin Lab Sci 2025;62:454-76. [Crossref] [PubMed]
39. Razaghizad A, Oulousian E, Randhawa VK, et al. Clinical Prediction Models for Heart Failure Hospitalization in Type 2 Diabetes: A Systematic Review and Meta-Analysis. J Am Heart Assoc 2022;11:e024833. [Crossref] [PubMed]
40. O'Neal J, Fencl JL. Strategies to Implement a Competency Assessment Verification Program. AORN J 2021;113:455-63. [Crossref] [PubMed]
41. Yoshida T, Hayakawa T, Kawadai T, et al. Implementation and Effectiveness of Measures to Prevent Recurrence of Incidents Developed through Root Cause Analysis Using Work Flowchart. Nihon Hoshasen Gijutsu Gakkai Zasshi 2024;80:304-10. [Crossref] [PubMed]
42. Fuhrman DY, Stanski NL, Krawczeski CD, et al. A proposed framework for advancing acute kidney injury risk stratification and diagnosis in children: a report from the 26th Acute Disease Quality Initiative (ADQI) conference. Pediatr Nephrol 2024;39:929-39. [Crossref] [PubMed]
43. Liu Q, Liu J, Lin B, et al. Novel Biomarkers in Membranous Nephropathy. Front Immunol 2022;13:845767. [Crossref] [PubMed]
44. Adam CA, Șalaru DL, Prisacariu C, et al. Novel Biomarkers of Atherosclerotic Vascular Disease-Latest Insights in the Research Field. Int J Mol Sci 2022;23:4998. [Crossref] [PubMed]
45. Wilnes B, Castello-Branco B, Branco BC, et al. Urinary L-FABP as an Early Biomarker for Pediatric Acute Kidney Injury Following Cardiac Surgery with Cardiopulmonary Bypass: A Systematic Review and Meta-Analysis. Int J Mol Sci 2024;25:4912. [Crossref] [PubMed]
46. Bojan M, Pieroni L, Semeraro M, et al. Cell-Cycle Arrest Biomarkers: Usefulness for Cardiac Surgery-Related Acute Kidney Injury in Neonates and Infants. Pediatr Crit Care Med 2020;21:563-70. [Crossref] [PubMed]
47. Chen JJ. Multicenter Observational Studies: Understanding the Basics of Data Sharing and Data User Agreements. J Neuroophthalmol 2024;44:1-4. [Crossref] [PubMed]
48. Hu M, Shi X, Song PX. Collaborative inference for treatment effect with distributed data-sharing management in multicenter studies. Stat Med 2024;43:2263-79. [Crossref] [PubMed]
49. Oh W, Nadkarni GN. Federated Learning in Health care Using Structured Medical Data. Adv Kidney Dis Health 2023;30:4-16. [Crossref] [PubMed]
50. Basu RK, Hackbarth R, Gillespie S, et al. Clinical phenotypes of acute kidney injury are associated with unique outcomes in critically ill septic children. Pediatr Res 2021;90:1031-8. [Crossref] [PubMed]
51. Wishahi M. Lower urinary tract dysfunction in pediatrics progress to kidney disease in adolescents: Toward precision medicine in treatment. World J Nephrol 2021;10:37-46. [Crossref] [PubMed]