Cheungpasitporn, Wisit

Artificial intelligence (AI)–based protein structure prediction has rapidly entered nephrology, providing plausible atomic models for channels, transporters, receptors and scaffolds that previously lacked structural information. AlphaFold and RoseTTAFold now underpin mechanistic interpretations of inherited kidney diseases, including autosomal-dominant polycystic kidney disease, cystinosis and Alport syndrome, and support experimental design in cryo-electron microscopy and cryo-electron tomography. For nephrologists, this means that renal proteins no longer need to be treated as black boxes: genetic, biochemical and physiological data can now be mapped onto concrete three-dimensional scaffolds. However, the apparent precision of AI models can be misleading. Renal proteins frequently fall within underrepresented classes, including multipass membrane proteins, intrinsically disordered scaffolds, and human-specific isoforms with shallow multiple sequence alignments, where confidence scores may not accurately reflect model reliability. Current predictors typically deliver static snapshots, whereas transporters, channels and slit-diaphragm complexes function as dynamic ensembles embedded in specific lipid and macromolecular environments. Attempts to use AlphaFold directly for variant classification or to estimate changes in thermodynamic stability or function show only modest agreement with mutagenesis data, particularly for gain-of-function and dominant-negative alleles that depend on subtle conformational and interaction changes. Common pitfalls include overinterpreting oligomeric stoichiometry inferred from AlphaFold Multimer, reading too much into domain orientations with high predicted aligned error, and dismissing low-confidence regions as modelling failures rather than recognising intrinsically disordered, potentially regulatory segments. Integrative approaches combining cryo-electron tomography, crosslinking mass spectrometry and molecular dynamics can mitigate these limitations but remain labour-intensive and low-throughput. An explainable, hypothesis-driven use of AI that treats predicted structures as working scaffolds to be cross-validated against imaging, biophysics and functional assays offers the most reliable path towards a genuinely mechanistic nephrology and responsible clinical translation of AI-based structural insights.

Rare diseases affect over 300 million individuals worldwide yet remain underdiagnosed and poorly characterized due to fragmented data, small cohorts, and phenotypic heterogeneity. Advances in artificial intelligence (AI) are enabling integration of genomics, imaging, electronic health records, and patient-generated data to support diagnosis, phenotyping, prognosis, and therapeutic discovery. In kidney care, these capabilities are reflected in tools for genomic variant prioritization, AI-assisted histopathology, and integrated risk stratification models for rare and complex kidney diseases. This review synthesizes current AI applications across the rare disease continuum and proposes a clinically grounded framework to distinguish exploratory models from systems that are methodologically robust and operationally deployable. We highlight advances that address data sparsity and heterogeneity, alongside persistent challenges in validation, generalizability, equity, and workflow integration. Finally, we outline future directions, including federated learning, digital twins, and AI-driven clinical decision agents, as pathways toward precision-guided, workflow-integrated rare disease care.

ABSTRACT Background Uremic toxins are a growing area of research in nephrology, with significant implications in the progression and treatment of chronic kidney disease (CKD) and the management of end‐stage kidney disease (ESKD). This bibliometric analysis aims to evaluate the global research trends, key contributors, and the impact of publications in the field of uremic toxins from 1991 to 2024. Methods The data were retrieved from the Science Citation Index Expanded (SCI‐EXPANDED) database. The study analyzed 2210 articles using various bibliometric indicators, including total publications ( TP ), average number of citations per publication ( CPP 2024 ), and the Y ‐index for author contributions. Publication trends were examined across countries, institutions, and authors. The analysis included document types, languages, Web of Science categories, and journals. Results The study revealed a steady increase in uremic toxin research publications. Japan, the USA, and China remained the most productive countries, with Japan maintaining the highest output, while Belgium achieved the highest citation impact. Citations of uremic toxin‐related research peaked 2 years following publication and plateaued after approximately 6 years. The research foci evolved, with CKD, ESKD, indoxyl sulfate, protein‐bound solutes, and gut microbiota gaining prominence. Studies on oxidative stress, inflammation, metabolomics, and advanced adsorption‐based toxin removal strategies increased markedly. International collaboration and interdisciplinary integration across nephrology, microbiology, and biochemistry correlated with higher citation performance and research visibility. Conclusion This bibliometric study provides an in‐depth analysis of uremic toxin research trends, identifies major contributors, and suggests future research priorities. The findings can inform future research directions, guide funding allocation, and strengthen global collaborative efforts towards improved management of toxin‐related kidney disease.

ABSTRACT Bayesian statistics is gaining momentum in biomedical research, offering a flexible and intuitive framework for integrating prior knowledge, managing uncertainty, and informing decision-making. In nephrology, a field marked by complex pathophysiology, small patient populations, and evolving therapeutic landscapes, Bayesian approaches may provide critical advantages over traditional statistical methods. Bayesian methodologies may help reshape nephrology by improving trial efficiency, refining diagnostic and prognostic precision, and strengthening causal inference. We conducted a comprehensive narrative review of peer-reviewed literature on Bayesian applications in nephrology, highlighting studies in adaptive clinical trial designs, dynamic risk prediction models, network meta-analyses, pharmacokinetic modeling, and Bayesian Mendelian randomization approaches, including Bayesian model averaging frameworks. Bayesian adaptive designs enabled more efficient and ethically sound trials, such as the WIndow-of-opportunity clinical trial platform for evaluation of novel treatment strategies in REnal cell cancer (WIRE) platform study in renal cell carcinoma. In diagnostics and prognostics, Bayesian models allowed individualized inference from dynamic data, as demonstrated in estimated glomerular filtration rate trajectory analyses and mortality prediction. Bayesian network meta-analyses improved comparative effectiveness research by incorporating probabilistic treatment rankings, while Bayesian frameworks strengthened causal inference through methods like Mendelian randomization (MR) -Bayesian Model Averaging and Bayesian Kernel Machine Regression. Bayesian approaches have emerging applications in nephrology, promoting a probabilistic framework that enhances the interpretation of complex clinical and research data and supports more individualized patient care. Their integration into both research and clinical practice requires investment in education and training, interdisciplinary collaboration, and the development of accessible analytical tools. Embracing Bayesian methods is likely to play an important role in advancing precision medicine and improving patient outcomes in nephrology.

Background: Artificial intelligence in nephrology has largely focused on predictive models for outcomes such as acute kidney injury (AKI), chronic kidney disease (CKD) progression, and transplant complications. Although these models demonstrate technical performance, their real-world clinical impact has remained limited because prediction alone rarely translates into coordinated clinical action. Clinical artificial intelligence agents represent workflow-native systems that operate in real time, interact bidirectionally with clinical environments, adapt to evolving patient and workflow states, and support coordinated clinical action rather than generating isolated predictions. This review proposes clinical artificial intelligence agents as a new paradigm for integrating artificial intelligence directly into nephrology workflows. Methods: We conducted a narrative synthesis of emerging literature on artificial intelligence systems, agentic artificial intelligence architectures, clinical decision support, and digital health infrastructures relevant to kidney care. Drawing from interdisciplinary sources in medicine, health informatics, and artificial intelligence research, we developed a conceptual framework describing the architecture, governance requirements, and evaluation principles of clinical artificial intelligence agents in nephrology. Results: Clinical artificial intelligence agents represent workflow-integrated systems capable of continuously perceiving patient data, reasoning under clinical constraints, planning tasks, and supporting coordinated clinical actions over time. We describe a layered architecture consisting of perception, cognition, planning and control, action, and learning components. Potential applications span the nephrology care continuum, including CKD management, AKI monitoring, dialysis and continuous renal replacement therapy (CRRT) optimization, kidney transplantation care coordination, glomerulonephritis management, and supervised patient-facing systems. Conclusions: Clinical artificial intelligence agents shift the role of artificial intelligence from isolated prediction toward longitudinal clinical orchestration. Future evaluation should prioritize workflow integration, time-to-action, clinician oversight, safety, and patient-centered outcomes rather than relying solely on traditional model performance metrics. This roadmap provides a conceptual foundation for the responsible development and clinical integration of agentic artificial intelligence systems in nephrology.

Background Kidney transplantation offers substantial clinical and economic benefits for patients with end-stage kidney disease (ESKD), yet organ shortages persist. Enhancing public awareness and health literacy regarding kidney donation is essential for effective donor recruitment. While online patient education materials are primary drivers of public perception, their readability, digital engagement, and accessibility remain underexplored barriers. Methods We analyzed the most prominent kidney donation websites identified through U.S.-based Google searches, including 11 primary organizations and 16 affiliated subdomains, for a total of 27 websites. Readability was benchmarked using Flesch-Kincaid, SMOG, and Gunning Fog indices. To assess qualitative metrics, we deployed a generative AI framework utilizing a Large Language Model (Claude) to conduct automated sentiment analysis and tone evaluation, validated by human review. We systematically mapped digital engagement features, including multimedia, interactive tools, and multilingual support, to determine content comprehensiveness. Results Websites consistently provided accurate information with a generally positive or neutral tone. Average readability exceeded recommended levels, with a combined mean grade of 12.3; 34% of websites were written at a college-level reading standard. Consensus across content was high. Multimedia elements were widely used, but engagement features were limited; only 30% of sites included extensive testimonials, and interactive tools were absent. AI-based analysis enabled standardized and reproducible evaluation, highlighting opportunities to improve accessibility, tone, and inclusivity. Conclusion Current U.S. kidney donation digital resources present a barrier to health equity due to excessive reading complexity and static engagement models. AI provides a scalable, reproducible framework to audit and optimize patient education materials. Future initiatives must leverage AI-guided content optimization to bridge the literacy gap, potentially increasing donor registration and access to transplantation.

Key Points Both sodium zirconium cyclosilicate and patiromer effectively lower serum potassium and provide clinical benefits. Sodium zirconium cyclosilicate is associated with higher risk of edema. Patiromer is associated with the increased risk of constipation and hypokalemia. Background Sodium zirconium cyclosilicate (SZC) and patiromer have recently been approved for the treatment of hyperkalemia. However, their efficacy in lowering serum potassium according to the phase of intervention, safety profile, and impact clinical outcomes remains to be fully defined. This study aims to assess their efficacy and tolerability in the management of hyperkalemia. Methods A comprehensive search of PubMed, Scopus, and Cochrane Central Register of Controlled Trials was conducted up to June 6, 2025. Eligible studies included randomized controlled trials reporting the effects of SZC and patiromer on laboratory parameters, clinical outcomes, and safety profiles. Meta-analyses were synthesized using a random-effects model, with results presented as weighted mean differences (WMD) or relative risks (RR), along with 95% confidence intervals (CI). Results Nineteen randomized controlled trials involving 3627 patients (1911 on SZC; 1716 on patiromer) were analyzed. During the initial 48–72 hours, only SZC significantly reduced serum potassium levels (WMD, −0.45 mEq/L; 95% CI, –0.61 to –0.29). At the end of treatment, both agents effectively lowered serum potassium (SZC: WMD, –0.50 mEq/L; 95% CI, –0.63 to –0.38; patiromer: WMD, –0.36 mEq/L; 95% CI, –0.51 to –0.21). SZC significantly normalized serum potassium and reduced hyperkalemia incidence, while patiromer enabled renin-angiotensin-aldosterone system inhibitor optimization by supporting target dosing and reducing discontinuation due to hyperkalemia. However, SZC was associated with a higher incidence of edema (RR, 2.92; 95% CI, 1.08 to 7.90) and patiromer with increased rates of constipation (RR, 2.96; 95% CI, 1.21 to 7.22) and hypokalemia (RR, 1.46; 95% CI, 1.05 to 2.03). Conclusions Both SZC and patiromer effectively lower serum potassium and provide clinical benefits, with distinct safety considerations. Therapy selection should be guided by treatment goals and the patient's individual risk profile.

ABSTRACT Background Acute kidney injury (AKI) in hospitalised children is a major complication associated with significant morbidity and mortality. The integration of artificial intelligence (AI)/machine learning (ML) models may enable early detection and risk stratification. This systematic review evaluates the performance of AI/ML models for predicting paediatric AKI across clinical settings. Methods We systematically searched PubMed, Embase, and Web of Science for studies applying AI/ML models to predict AKI in paediatric populations. Studies reporting performance metrics such as the area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and F1 score were included. Results Among 470 records identified, 11 studies met the inclusion criteria, with 14 AI/ML models used. The overall sample size included 33 949 paediatric patients with an AKI proportion of 12.5%. Meta-analyses of the AUC were conducted on neural network, gradient boosting, and logistic regression. Gradient boosting had the highest pooled AUC of 0.873 (95% confidence interval 0.836–0.909). Random forest demonstrated the highest median sensitivity (0.821), specificity (0.942), PPV (0.860), NPV (0.935), and accuracy (0.821); however, these metrics could not be pooled due to inconsistent reporting and limited validation. Conclusion Gradient boosting, random forest, and logistic regression demonstrated reasonable predictive performance for paediatric AKI prediction within specific clinical contexts. However, small sample size, heterogeneity, lack of testing/validation cohorts, insufficient data, and inconsistent patient populations and AKI diagnostic criteria restrict generalisability.

Key Points Baseline kidney function and chronic structural injury strongly predicted both renal recovery and long-term clinical outcomes. Tubulointerstitial nephritis was the predominant lesion in Sjögren syndrome with renal involvement. Early detection and management of renal involvement in Sjögren syndrome are critical to preventing irreversible kidney damage. Background Renal involvement in primary Sjögren syndrome (SS) is uncommon but clinically consequential. Previous studies have been limited by small samples and incomplete biopsy data. We evaluated a large biopsy-confirmed cohort to characterize clinicopathologic features of SS with renal involvement and identify predictors of renal recovery and long-term outcomes. Methods We retrospectively identified adults with SS and kidney biopsy at Mayo Clinic (2012–2025). Clinical, laboratory, and histologic data were extracted. Predictors of complete renal recovery within 6 months, defined as serum creatinine (sCr) returning to within 25% of baseline or <1.4 mg/dl if baseline was unknown, were evaluated using logistic regression. Cox regression assessed long-term risk of a composite end point: newly developed or progressive CKD, ESKD, or all-cause mortality. Results Fifty-six patients were included (median age 57 years; 91% female). Extraglandular manifestations occurred in 76% and baseline CKD in 65%. The median sCr at baseline and biopsy was 1.3 and 1.6 mg/dl, respectively. Anti-Sjogren's syndrome antigen A, often linked with Ro ribonucleoprotein was positive in 80%. Low C3 occurred in 14% and low C4 in 25%. Tubulointerstitial nephritis was the predominant lesion (71%). Moderate/severe interstitial fibrosis/tubular atrophy and arteriosclerosis were present in 36% and 41%, respectively. Immunosuppressive therapy was applied in 84%. Complete recovery occurred in 67%. Over a median 3.9 years (interquartile range, 1.4–7.2), 43% reached the composite end point. Baseline sCr ≥2 mg/dl, 24-hour proteinuria ≥1 g, and presence of segmental glomerulosclerosis were associated with lower odds of recovery (odds ratio=0.16, 0.13 and 0.11, P = 0.02, 0.001 and 0.008, respectively) and higher long-term risk (hazard ratio=4.51, 2.92 and 3.52; P = 0.002, 0.01 and 0.02, respectively). Moderate/severe arteriosclerosis also increased long-term risk (hazard ratio=2.56; P = 0.03). Conclusions In biopsy-confirmed SS-related renal involvement, the severity of kidney dysfunction and the extent of chronic glomerular and vascular injury strongly predict renal prognosis. Early detection and targeted management of high-risk features may improve long-term renal outcomes.

Hemodiafiltration (HDF) is increasingly used because of its enhanced theoretical clearance of diverse uremic toxins, particularly middle molecules and inflammatory cytokines, relative to conventional hemodialysis (HD), yet evidence on its biochemical benefits remains conflicting. Therefore, this meta-analysis was performed to evaluate the effects of HDF versus HD on uremic toxins, inflammation, anemia, and nutritional parameters. A systematic literature search was conducted using PubMed, Scopus, and the Cochrane Central Register of Controlled Trials to identify relevant studies. Only randomized controlled trials (RCTs) were included. Random-effects meta-analyses were performed to evaluate changes in the prespecified outcomes. Twenty-four RCTs involving 6072 dialysis patients were included. Compared with conventional HD, HDF was associated with significant reductions in serum phosphorus (weighted mean difference [WMD] −0.28 mg/dL; 95% CI −0.44 to −0.12) and β2-microglobulin (WMD −4.84 mg/dL; 95% CI −6.13 to −3.54). HDF also significantly reduced serum urea and C-reactive protein (CRP) levels, along with weekly erythropoietin requirements. Serum albumin levels were slightly but significantly lower in the HDF group than in the conventional HD group (WMD –0.06 g/dL; 95% CI −0.10 to −0.01); however, the clinical significance of such a difference remains uncertain. Higher convective volumes were identified as a key determinant of greater reductions in β2-microglobulin and CRP. Compared with conventional HD, HDF demonstrated superior reductions in several surrogate endpoints, including serum phosphorus, urea, β2-microglobulin, CRP, and weekly erythropoietin requirements. Reduced need for phosphate binders and anemia management may lower treatment-related costs.