Acute kidney injury prediction and prognostication using machine learning

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Not quantitatively reported — described qualitatively as 'high predictive accuracy'; no pooled AUROC, sensitivity, specificity, or hazard ratios provided

• Primary Outcome: Predictive accuracy of AI/ML models for AKI onset and clinical outcomes across ICU, sepsis, postoperative, and post-contrast clinical settings

• Nnt Or Sensitivity: Not reported — no NNT, sensitivity, specificity, positive predictive value, or negative predictive value estimates are provided in the abstract or available metadata

• Confidence Interval: Not reported — no statistical synthesis performed

Clinical Application

Clinical implementation of ML-based AKI prediction tools requires integration with electronic health record (EHR) systems, real-time data pipelines, and clinical decision support infrastructure. The review acknowledges the need for user-friendly platforms. Significant barriers remain including model interpretability (black-box concerns), workflow integration, clinician training, and ongoing model maintenance to prevent performance drift. Prospective randomised trials demonstrating that ML-triggered alerts improve patient outcomes — rather than merely predicting AKI — are largely absent from the current literature. In Australia, AKI management is guided by KDIGO 2012 criteria, which remain the clinical standard. The Australian Commission on Safety and Quality in Health Care (ACSQHC) has identified AKI as a priority safety concern. Several Australian tertiary centres have piloted EHR-integrated AKI alerting systems, though ML-based tools are not yet standard of care. The TGA would require regulatory approval for any ML-based clinical decision support tool classified as a medical device under the Software as a Medical Device (SaMD) framework. PBS implications are indirect — earlier AKI detection could reduce dialysis initiation rates and associated costs. RACGP and ANZSN (Australian and New Zealand Society of Nephrology) guidelines do not currently endorse specific ML tools for AKI prediction. Australian implementation would also need to address Indigenous health equity, as Aboriginal and Torres Strait Islander peoples have disproportionately high rates of CKD and AKI, and training data from predominantly non-Indigenous cohorts may not generalise to this population. Patients at risk of AKI in acute care settings — particularly ICU patients, those with sepsis, post-surgical patients, and patients receiving iodinated contrast media. The review does not specify age ranges, baseline renal function thresholds, or comorbidity profiles that would define the optimal target population for ML-based prediction tools.

Abstract

Acute Kidney Injury (AKI) is a major health concern with high costs and poor outcomes, partly due to late diagnosis. This paper reviews the application of artificial intelligence (AI) and machine learning (ML) to overcome these limitations by enabling earlier and more accurate prediction and prognostication of AKI. Many studies on AI and ML models used to predict and prognosticate AKI were included in the review. The focus was on models that analyze complex datasets, including real-time data streams like novel biomarkers and continuous vital signs, to achieve earlier and more accurate predictions than conventional methods. ML models show high predictive accuracy for AKI onset and outcomes across various clinical settings, including intensive care units, sepsis, and postoperative and postcontrast situations, with key findings like the successful integration of real-time data to reflect the evolving nature of kidney injury. AI and ML offer a powerful, proactive solution for AKI management. By leveraging diverse data, these technologies can significantly improve patient outcomes and reduce healthcare costs. While their promising performance warrants further exploration, successful clinical integration will require user-friendly platforms and continued validation.

References

1. 1.Senatore, A., Fiorentino, M., Bonerba, B., Baldassini Moraes, B., Ciocchetti, P., Grandaliano, G., & Pesce, F. (2026). Acute kidney injury prediction and prognostication using machine learning. International Urology and Nephrology. https://doi.org/10.1016/j.ebiom.2024.105726