Ghazi, Lama

Scientific evidence regarding AKI and acute kidney care has advanced in recent years. However, a knowledge gap remains on the implementation of these evidence-based practices (EBPs) into clinical care. Implementation science (IS) is focused on ensuring that this knowledge is translated in an effective, efficient and sustainable fashion. Objectives: 1) Define the current status of IS; 2) Define a roadmap for accelerating IS with a focus on patient/care partner advocacy, digital tool application, social determinants of health, and resource-limited settings; and 3) Develop a robust and broad research agenda incorporating IS methodology into the programs. Evidence Review: ADQI XXXV was conducted through a modified Delphi process with virtual meetings preceding an in-person meeting. Five workgroups were determined a priori to focus on 1) IS definitions and methods applicable to AKI and acute kidney care; 2) IS literature in AKI and acute kidney care; 3) Innovations in IS to accelerate the adoption, adherence to, and sustainment of EBPs in AKI and acute kidney care; 4) IS methods in resource-limited settings; and 5) Recommendations to identify and evaluate EBPs that are ready for implementation or de-implementation. Prior to the in-person meeting, each workgroup met virtually to review the literature and develop framework questions to address their objectives. During the two-day in person meeting, through iterative discussions, questions and supporting statements were finalized. These questions and statements were agreed upon through voting to achieve consensus, defined as agreement of ≥80%. Conclusions and Relevance: We report a structured multidisciplinary consensus for defining the role of IS in AKI and acute kidney care. Future programs should address these consensus questions and apply these statements along with IS methodology in the translation of science into clinical practice and the implementation/de-implementation of EBPs in clinical care.

ABSTRACT Artificial intelligence (AI), encompassing machine learning, deep learning and generative AI, is poised to redefine nephrology by enabling earlier detection, more precise risk stratification and workflow-integrated clinical decision support across the spectrum of kidney disease. This state-of-the-art review synthesizes emerging applications of AI in acute kidney injury (AKI), chronic kidney disease (CKD), dialysis and kidney transplantation, with attention to clinical integration, real-world validation, workflow implementation and translational challenges. In AKI, predictive models trained on high-frequency electronic health record data and intensive care unit telemetry have demonstrated strong performance in forecasting critical events, yet translation into routine clinical workflows remains limited. In CKD, machine learning tools support progression risk stratification and phenotype clustering, with the potential to inform individualized surveillance and therapy. AI-enabled dialysis management systems optimize ultrafiltration, anemia control and vascular access surveillance, while generative AI and large language models offer novel capabilities for clinical documentation, triage and patient education. In transplantation, AI applications span organ allocation, dynamic graft monitoring and digital pathology-assisted rejection classification, with validated tools such as the iBox system gaining regulatory recognition. Implementation challenges include data heterogeneity, bias, interpretability, regulatory uncertainty and workflow integration. Looking ahead, multimodal integration of imaging, pathology and multi-omics data may support biologically informed precision nephrology. Reinforcement learning, digital twins and ambient intelligence are emerging as adaptive decision-support paradigms rather than autonomous care systems. Regulatory frameworks are evolving to accommodate adaptive algorithms, underscoring the need for clinician engagement in model development, validation and deployment. As AI matures from pilot to practice, nephrologists who embrace and help shape these tools will lead the transition toward a more personalized, efficient and equitable future for kidney care.

Key Points Pooled discrimination metrics were acceptable (area under the receiver operating characteristic curve >0.70) for all AKI risk classification categories in both internal and external validation. Better performance was observed in most recently published studies and those with a low or unclear risk of bias. Significant heterogeneity in patient populations, definitions, clinical predictors, and methods limit implementation in real-world clinical scenarios. Background Artificial intelligence through machine learning models seems to provide accurate and precise AKI risk classification in some clinical settings, but their performance and implementation in real-world settings has not been established. Methods PubMed, Excerpta Medica (EMBASE) database, Web of Science, and Scopus were searched until August 2023. Articles reporting on externally validated models for prediction of AKI onset, AKI severity, and post-AKI complications in hospitalized adult and pediatric patients were searched using text words related to AKI, artificial intelligence, and machine learning. Two independent reviewers screened article titles, abstracts, and full texts. Areas under the receiver operating characteristic curves (AUCs) were used to compare model discrimination and pooled using a random-effects model. Results Of the 4816 articles initially identified and screened, 95 were included, representing 3.8 million admissions. The Kidney Disease Improving Global Outcomes (KDIGO)-AKI criteria were most frequently used to define AKI (72%). We identified 302 models, with the most common being logistic regression (37%), neural networks (10%), random forest (9%), and eXtreme gradient boosting (9%). The most frequently reported predictors of hospitalized incident AKI were age, sex, diabetes, serum creatinine, and hemoglobin. The pooled AUCs for AKI onset were 0.82 (95% confidence interval, 0.80 to 0.84) and 0.78 (95% confidence interval, 0.76 to 0.80) for internal and external validation, respectively. Pooled AUCs across multiple clinical settings, AKI severities, and post-AKI complications ranged from 0.78 to 0.87 for internal validation and 0.73 to 0.84 for external validation. Although data were limited, results in the pediatric population aligned with those observed in adults. Between-study heterogeneity was high for all outcomes (I 2 >90%), and most studies presented high risk of bias (86%) according to the Prediction Model Risk of Bias Assessment Tool. Conclusions Most externally validated models performed well in predicting AKI onset, AKI severity, and post-AKI complications in hospitalized adult and pediatric populations. However, heterogeneity in clinical settings, study populations, and predictors limits their generalizability and implementation at the bedside.

It remains uncertain whether intensive control of blood pressure (BP) results in a lower risk of atrial fibrillation (AF) in patients with hypertension. Using data from SPRINT (Systolic Blood Pressure Intervention Trial), which enrolled participants with hypertension at increased risk of cardiovascular disease, we examined whether intensive BP lowering (target systolic BP [SBP] <120 mm Hg), compared with standard BP lowering (target SBP<140 mm Hg), results in a lower risk of AF. This analysis included 8022 participants (4003 randomized to the intensive arm and 4019 to standard BP arm) who were free of AF at the time of enrollment and with available baseline and follow-up electrocardiographic data. AF was ascertained from standard 12-lead electrocardiograms recorded at biannual study examinations and an exit visit. During up to 5.2 years of follow-up and a total of 28 322 person-years, 206 incident AF cases occurred; 88 in the intensive BP-lowering arm and 118 in the standard BP-lowering arm. Intensive BP lowering was associated with a 26% lower risk of developing new AF (hazard ratio, 0.74 [95% CI, 0.56–0.98]; P =0.037). This effect was consistent among prespecified subgroups of SPRINT participants stratified by age, sex, race, SBP tertiles, prior cardiovascular disease, and prior chronic kidney disease when interactions between treatment effect and these subgroups were assessed using Hommel adjusted P values. In conclusion, intensive treatment to a target of SBP <120 mm Hg in patients with hypertension at high risk of cardiovascular disease has the potential to reduce the risk of AF. Registration— URL: https://www.clinicaltrials.gov ; Unique identifier: NCT01206062.

Background Blood pressure ( BP ) varies over time within individual patients and across different BP measurement techniques. The effect of different BP targets on concordance between BP measurements is unknown. The goals of this analysis are to evaluate concordance between (1) clinic and ambulatory BP , (2) clinic visit‐to‐visit variability and ambulatory BP variability, and (3) first and second ambulatory BP and to evaluate whether different clinic targets affect these relationships. Methods and Results The SPRINT (Systolic Blood Pressure Intervention Trial) ambulatory BP monitoring ancillary study obtained ambulatory BP readings in 897 participants at the 27‐month follow‐up visit and obtained a second reading in 203 participants 293±84 days afterward. There was considerable lack of agreement between clinic and daytime ambulatory systolic BP with wide limits of agreement in Bland‐Altman plots of −21 to 34 mm Hg in the intensive‐treatment group and −26 to 32 mm Hg in the standard‐treatment group. Overall, there was poor agreement between clinic visit‐to‐visit variability and ambulatory BP variability with correlation coefficients for systolic and diastolic BP all <0.16. We observed a high correlation between first and second ambulatory BP ; however, the limits of agreement were wide in both the intensive group (−27 to 21 mm Hg) and the standard group (−23 to 20 mm Hg). Conclusions We found low concordance in BP and BP variability between clinic and ambulatory BP and second ambulatory BP . Results did not differ by treatment arm. These results reinforce the need for multiple BP measurements before clinical decision making.

Blood pressure (BP) varies over time within individual patients and across different BP measurement techniques. The effect of different BP targets on the concordance between BP measurements is unknown. The goal of this analysis was to evaluate concordance in: 1) clinic BP and ambulatory BP, 2) clinic visit-to-visit variability and ambulatory BP variability and 3) initial and repeat ambulatory BP. We also sought to evaluate whether treatment assignment of intensive vs standard BP target affected these relationships. The Systolic Blood Pressure Intervention Trial (SPRINT) ambulatory blood pressure monitoring ancillary study obtained ambulatory BP readings in 897 SPRINT participants at the 27 month follow up visit and 203 consecutive repeat ambulatory BP readings taken an average of 9.8 months later. There was poor agreement between clinic systolic BP and daytime ambulatory systolic BP (limits of agreement in Bland-Altman plots of -21 to 34 mm Hg in the intensive treatment group and -26 to 32 mm Hg in the standard treatment group). There was poor agreement between clinic visit-to-visit variability (coefficient of variation) and ambulatory BP variability (coefficient of variation of a 24 hr ambulatory BP) with correlation coefficients for systolic BP <0.16. While there was a high correlation between ambulatory BP at 27 months and repeat ambulatory BP (r ~0.56), there was significant variability between repeat ambulatory BP assessments (limits of agreement of -27 to 21 mm Hg in the intensive group and -23 to 20 mm Hg in the standard group). In conclusion, irrespective of treatment target, we found low concordance in BP and BP variability between clinic BP and ambulatory BP, and additionally between repeat ambulatory BP assessments. These results reinforce the need for multiple guideline adherent BP measurements prior to clinical decision making.