Sejoong Kim

We aimed to develop a transformer-based model to predict kidney outcomes from longitudinal electronic health records (EHR) of patients with diabetes and normal kidney function, and compared it with. logistic regression (LR), support vector machine (SVM), LightGBM, and XGBoost. Using EHR data from 8,772 patients (Seoul National University Bundang Hospital, 2003–2014), we predicted ≥ 30% estimated Glomerular filtration rate (eGFR) decline and advanced chronic kidney disease (CKD) stage at 1, 3, 5, and 10 years (event 1.2–8.4%). ForeGFR decline, LightGBM and XGBoost both outperformed the transformer at every horizon (AUROC 0.68–0.80 vs. 0.56–0.75), for advanced CKD, the transformer-based model achieved AUROC of 0.77–0.85, comparable to the trees. At a matched operating point (recall 0.8), F1 scores were comparable across models. The transformer was overconfident and, unlike the trees, did not exceed treat-all net benefit in decision-curve analysis. It was, however, more robust to input degradation: retaining only 25% of laboratory values reduced AUROC by 0.03 versus 0.23–0.30 for the trees. Under temporal validation (training ≤ 2009, testing 2009–2016), AUROC fell by an average of 0.09. These findings support the technical feasibility of sequence models for EHR data while highlighting the need for recalibration and external validation before clinical use.

Background: Albuminuria is one of the factors promoting the progression of chronic kidney disease (CKD). The study aimed to assess the efficacy and safety of calcitriol for the reduction of microalbuminuria in patients with nondiabetic nephropathy.Methods: In this randomized, double-blind, placebo-controlled, and multicenter study, adult patients with nondiabetic CKD stage 3 or greater and albuminuria were included. Participants were administered calcitriol or placebo for 6 months and followed for up to 12 months. The primary outcome was the change in urine protein-to-creatinine ratio (UPCR), and secondary outcomes included the changes in renal function and vitamin D level. The safety was assessed by recording adverse events during the treatment and follow-up.Results: A total of 159 subjects were enrolled. The UPCR at 24 and 48 weeks was significantly decreased compared to the baseline in the calcitriol group (ΔUPCR, –0.24 g/g [95% CI, –0.43 to –0.05] and –0.22 g/g [95% CI, –0.43 to –0.01], respectively), but the mean changes of UPCR during 24 weeks and 48 weeks were no significant difference between the two groups. No significant differences were in the change in renal function and vitamin D level. Seventy-eight adverse events were reported during the treatment phase, and there were no significant differences in the type or frequency of adverse events between the two groups.Conclusion: Although calcitriol treatment showed a significant reduction of proteinuria from baseline, the effect was insufficient in nondiabetic CKD compared to placebo. Therefore, the use of calcitriol for the reduction of albuminuria is worth considering.

Although blood pressure and lipid control are key in advanced chronic kidney disease (CKD), the optimal targets for systolic blood pressure (SBP) and low-density lipoprotein cholesterol (LDL-C) remain unclear because patients with estimated glomerular filtration rate (eGFR) <45 mL/min/1.73 m2 were often excluded or underrepresented in prior trials. This study aims to compare the clinical outcomes of intensive versus standard control strategies for SBP and LDL-C in patients with advanced CKD. This multicenter, open-label, randomized controlled trial with a 2 × 2 factorial design is being conducted across 13 centers in Korea. Adults aged ≥19 years with eGFR ≥15 and <45 mL/min/1.73 m2 , SBP ≥130 mmHg, and LDL-C ≥100 mg/dL are eligible. Participants will be randomized according to intensive or standard targets for SBP (<120 mmHg vs. <140 mmHg) and LDL-C (<70 mg/dL vs. <100 mg/dL). The effects of blood pressure and lipid interventions will be evaluated independently, sharing the same primary endpoint: time to the first composite kidney outcome, including ≥40% sustained decline in eGFR, initiation of kidney replacement therapy, sustained eGFR <10 mL/min/1.73 m2 , or kidney death. Secondary endpoints will include eGFR slope and major adverse cardiovascular events. Participants will be followed up for 3 years, with an additional 3 years for the observational phase. This report describes the rationale and design; trial results are pending. As the first randomized trial to determine whether intensive SBP and LDL-C lowering improve kidney outcomes in advanced CKD, this study could help define optimal targets and inform future guideline recommendations.

Background: Real-time prediction of intradialytic hypotension (IDH) using deep learning has shown high accuracy; however, existing models typically treat all IDH events equally, without distinguishing between initial and recurrent occurrences within a dialysis session. This conventional approach neglects the distinct underlying physiological mechanisms and clinical intervention requirements of each occurrence type. This study aimed to systematically examine IDH recurrence patterns to evaluate their impact on model performance and to identify methodological improvements. Methods: We retrospectively analyzed 12,767 hemodialysis sessions from 66 patients. Recurrent IDH was defined as an event occurring ≥30 min after the initial IDH. Deep learning models, including ConvMixer, temporal convolutional network, and long short-term memory with attention, were compared with a rule-based naïve baseline that predicted IDH solely from prior occurrence. Modeling strategies explicitly incorporating recurrence information were implemented. Model robustness across systolic blood pressure subgroups was evaluated and enhanced using adversarial training. Results: The probability of IDH increased markedly from 0.7–10.4% before initial events to 11.7–65.7% thereafter. Conventional evaluation that aggregated all IDH events overestimated performance, with particularly large gaps in F1 score and AUPRC between initial and recurrent IDH predictions. The naïve baseline achieved an area under the receiver operating characteristic (AUROC) curve of 0.798 without training, highlighting the strong influence of recurrence patterns on predictive performance. Incorporating recurrence-aware features and loss weighting improved AUROC by up to 8.0 percentage points across architectures. Adversarial training further reduced subgroup disparities while preserving overall model performance. Conclusions: Incorporating recurrence patterns into IDH prediction models demonstrated improvements in accuracy, robustness, and comparability across studies. We recommend standardized evaluation protocols that explicitly account for recurrence to enhance clinical applicability and reliability.

Acute kidney injury (AKI) is a common and serious disease, needing a comprehensive and management-based approach. While traditional practice focused on management after occurrence, recent practice is expanding its focus to prevention and novel therapeutic strategies, using new discoveries in AKI research. New biomarkers have been approved along with therapeutic agents, which target different pathophysiological pathways of AKI. In this review, we first summarize the current understanding of AKI definition and epidemiology, including global disparities and recent changes related to pandemics. Next, we highlight recent research directions regarding biomarkers, therapeutic agents, continuous renal replacement therapy, digital healthcare and artificial intelligence, and global collaborations. The aim of this review is to provide a comprehensive view of the direction in which AKI research is heading and link these areas to changes that are happening in clinical practice, with a goal to push innovative research accomplishments.

Hyponatremia undercorrection (i.e., nonoptimal correction) is insufficient to resolve cerebral edema and is associated with poor clinical outcomes. This study aimed to identify risk factors of undercorrection under controlled hypertonic saline treatment. This post-hoc analysis of the prospective randomized controlled SALSA (Efficacy and Safety of Rapid Intermittent Correction Compared With Slow Continuous Correction With Hypertonic Saline) trial included 172 patients with glucose-corrected serum sodium (sNa) ≤125 mmol/L after excluding six without follow-up sNa measurements. Undercorrection was defined as sNa increases of <5 or <10 mmol/L within 24 or 24-48 hours, respectively, or failure to reach 130 mmol/L within 24-48 hours. The mean age was 72.8 years; the mean sNa level was 118.2 mmol/L. Undercorrection occurred in 15.1% of patients (26/172). These patients received more hypertonic saline (932 mL vs. 486 mL, p < 0.001) and had lower urine output over 48 hours (2,647 mL vs. 4,117 mL, p = 0.004). Risk factors included higher urine osmolality and serum calcium and creatinine levels, and lower body mass index, systolic blood pressure, white blood cell count, and albumin levels. Although most etiologic categories were not independently associated with undercorrection, thiazide use was inversely associated with undercorrection in the multivariable analysis. Undercorrection did not increase mortality but was associated with prolonged hospitalization (19.6 days vs. 7.8 days, p = 0.01). Despite protocol-based hypertonic saline treatment, undercorrection occurred in 15% of patients and was associated with patient-specific risk factors. These findings emphasize the need for individualized sodium correction strategies to optimize patient outcomes.

Background: Membranous nephropathy (MN) is a specific autoimmune disease affecting kidneys. It is characterized by the accumulation of immune complexes in the glomerular basement membrane. Renal biopsy is currently the standard procedure to confirm the diagnosis, although the presence of autoantibodies against the phospholipase A2 receptor (PLA2R) can also help diagnose. In this study, we aimed to investigate the potential of urinary exosomes as noninvasive markers for diagnosing MN.Methods: Exosomes were extracted from urine samples of five patients with MN and four healthy controls. The concentration of PLA2R was measured in both urine and isolated exosomes using enzyme-linked immunosorbent assay techniques. The measurements were adjusted based on the urine creatinine (UCr) level of each participant.Results: The levels of PLA2R/UCr were investigated in urine and urine-derived exosomes from patients and controls. Results of the analysis revealed significantly higher expression of PLA2R/UCr in patients compared to the control group (p < 0.05). Furthermore, the expression level of PLA2R/UCr was higher in urine-derived exosomes than in urine samples. Additionally, a positive correlation was observed between the expression levels of PLA2R/UCr and the urine protein-to-creatinine ratio, with urine-derived exosomes exhibiting a stronger correlation than urine samples.Conclusion: Studies have indicated that measuring exosomal PLA2R/UCr levels in urine could be a noninvasive method for diagnosing MN. Using urine-derived exosomes could also reduce the burden of performing a biopsy on patients and facilitate follow-up treatment, such as monitoring for future recurrence.

The Korean Society of Critical Care Nephrology, in collaboration with the Korean Society of Nephrology, has published a Clinical Practice Guideline (CPG) for Continuous Kidney Replacement Therapy (CKRT). This guideline was developed with the goal of improving patient outcomes and reducing complications, based on the latest research related to the application and prescription of CKRT. The development of this guideline involved multidisciplinary participation from leading experts in nephrology and critical care medicine, with methodological support for the guideline development process. This CPG consists of recommendations addressing 17 key questions related to the initiation of CKRT, dose, modality, dialysate, ultrafiltration, anticoagulation, application in special situations, and multidisciplinary care. Each recommendation begins with a statement graded by the strength of the recommendation and the quality of the supporting evidence, followed by a rationale that supports the recommendation. We hope that this CPG will offer practical guidance in clinical settings and contribute to the improvement of patient outcomes.