Abstract: FR-PO0033
Artificial Intelligence (AI) in Kidney Care: A Scoping Review of Prediction, Diagnosis, and Treatment Applications
Session Information
- Artificial Intelligence: Clinical Prediction, Decision Support, and Digital Kidney Care
October 23, 2026 | Location: Exhibit Hall A, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Neupane, Raghavee, Nova Southeastern University, Fort Lauderdale, Florida, United States
- Yetman, Hailey, Icahn School of Medicine at Mount Sinai, New York, New York, United States
- Chan, Lili, Icahn School of Medicine at Mount Sinai, New York, New York, United States
Background
Artificial intelligence is increasingly being integrated across nephrology. AI has shown potential to improve early disease detection, risk stratification, treatment optimization, and healthcare resource utilization. However, the overall scope, performance, and clinical applications of these tools remain incompletely characterized.
Methods
We conducted a review of PubMed, Web of Science, Ovid, and Embase for English-language primary studies from the past five years. 378 studies were screened. Studies were categorized by clinical domain, AI methodology, and dataset characteristics.
Results
AI applications were identified across AKI, CKD, dialysis, and transplantation, using retrospective EHR and registry datasets with ensemble machine learning (XGBoost, random forests) and DL models. Most studies utilized the layering of multiple models. Many trials assessed AI tools' performance in predicting treatment management and patient outcomes, including disease progression, mortality, and treatment efficacy. AI was used to recognize patterns in large, unstructured datasets to identify patient-level risk factors via prediction or clustering, as well as to analyze physiologic or multi-omics data associated with outcomes. Image-analysis tools were also applied to patient samples and medical imaging to support diagnosis and clinical care. Despite promising performance metrics in many models with AUC ranging from 0.66 to 0.99, clinical deployment remains limited due to challenges including sample size, external validation, model interpretability, and integration into clinical workflows.
Conclusion
AI is increasingly applied across all domains of nephrology, with particularly rapid expansion in dialysis-related predictive modeling and real-time complication detection. While performance is generally strong in retrospective studies, clinical translation is limited by a lack of prospective validation, external generalizability, and workflow integration.
Acknowledgment
We would like to thank Lili Chan for her mentorship, guidance, and invaluable support throughout this work.
The authors used ChatGPT (OpenAI) for grammar and language editing. The authors reviewed and take full responsibility for the final content.
Characteristics and Clinical Applications of AI Models in Nephrology
| Categories | # of papers | AI models | Outcomes Reported |
| AKI Prediction | 58 | Logistic regression, LASSO, XGReost LightGRA. random forest. | AKI severity prediction and RRT need. |
| AKI Prognosis | 24 | Logistic regression, XGBeast, random forest, SVM, KNN. | Mortality, persistent/stage 3 AKI, and CKD/ESKD progression. |
| CKD Prediction | 24 | XGBoost, tree-based ML models, deep learning models. | CKD onset, detection, and staging. |
| CKD Diagnosis and monitoring | 23 | Logistic regression, LASSO, random forest, gradient boosting, CNNs, multimodal models. | Renal fibrosis severity, CKD related complications (e.g., cardiovascular events, bone/mineral disorders, sarcopenia, cognitive impairment). |
| Dialysis | 31 | Logistic regression, random forest, XGBoost, SVM, KNN, neural networks, CNNs, LSTMs, transformers. | Dialysis complications, mortality, hospitalization, and diagnostic support. |
| Kidney Transplant | 29 | XGBoost, LightGBM, CatBoost, random forest, boosting, logistic regression, SVM, KNN. | Transplant outcomes, graft failure, and immunosuppression management. |