Chinese Journal of Blood Purification

12 April 2026, Volume 25 Issue 04 Previous Issue    Next Issue

Select

Chinese expert consensus on the insertion, replacement and removal of tunneled cuffed central venous catheters for hemodialysis (1st edition)  Chinese Expert Consensus Writing Group on the Insertion 2026, 25 (04):  265-277.  doi: 10.3969/j.issn.1671-4091.2026.04.001 Abstract ( 530 )   Central venous catheter is an important type of vascular access, among which tunneled cuffed central venous catheter (TCC) is increasingly used and has been adopted by many institutions in China. However, due to inconsistencies in concepts and techniques across different institutions, some demonstrate deficiencies in the indications for TCC insertion, replacement, and removal, as well as in procedural standardization and operational risk management. Given that TCC insertion and replacement are inherently high-risk procedures, non-standard practices may lead to complications such as asphyxia, catheter dysfunction, and catheter-related infections, which can severely affect patients' dialysis outcomes and quality of life. Meanwhile, many institutions still require the performance of TCC insertion, replacement, and removal. In view of this, an expert consensus panel established a writing committee to systematically outline standardized oper ating procedures for the insertion, replacement, and removal of tunneled cuffed central venous catheters, integrating evidence-based consensus with clinical experience. It is hoped that under the guidance of this consensus, the techniques for TCC insertion, replacement, and removal in China can be further promoted, standardized, and refined, thereby benefiting a broader population of hemodialysis patients. Metrics

Select

Comparative analysis of prediction models based on different algorithms in the prediction of the risk of death in the short term of maintenance hemodialysis for uremia HANG Miao-miao, LV Hong-hong 2026, 25 (04):  278-283.  doi: 10.3969/j.issn.1671-4091.2026.04.002 Abstract ( 123 )   Objective  This study aimed to compare different machine learning algorithms to identify the optimal prediction model for the short-term mortality risk in patients undergoing maintenance hemodialysis (MHD), for uremia, providing a basis for early clinical risk stratification and intervention. Methods  A total of 197 uremic patients undergoing MHD at The Second Affiliated Hospital of Air Force Military Medical University from April 2022 to May 2024 were retrospectively enrolled. Based on 12-month survival outcomes, they were categorized into a death group (n=54) and a survival group (n=143). Using independent risk factors identified by multivariate logistic regression, five machine learning models—Logistic Regression (LR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN)—were developed. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, F1 score, and accuracy. Decision Curve Analysis (DCA) was used to assesse clinical net benefit, and SHapley Additive exPlanations (SHAP) was applied to quantify the contribution of each clinical feature to the predictions. Results  Multivariate Logistic regression showed that diabetes (OR=14.347, 95%CI: 1.796～114.624, P=0.012), total cholesterol (OR=62.811, 95%CI: 2.463～1601.852, P=0.012), CRP (OR=4.723, 95% CI: 1.592-14.011, P=0.005) were independent risk factors for short-term mortality in patients with uremic MHD (P <0.05), while ALB (OR=-0.917, 95%CI: (0.231～0.693, P=0.001) was an independent protective factors (P<0.05). Comparison of the performance of the five algorithms reveals that the AUC value of the KNN model achieved the highest AUC value of 0.909, and an F1 score of 0.796, identifying it as the optimal predictive model for short-term mortality risk in uremic MHD patients. SHAP analysis identified ALB, CRP, diabetes, and total cholesterol as the most influential predictors. Conclusion: The KNN model integrates key indicators such as ALB, CRP, diabetes and total cholesterol, effectively predicts the short-term mortality risk in patients with uremic MHD. Metrics

Select

Maintenance hemodialysis patients with lower serum reverse triiodothyronine levels are independently associated with sarcopenia HE Xiao-jiao, LI Shuang, QIU Zhen-liang, LIU Yun, ZHONG Xiao-shi 2026, 25 (04):  284-0.  doi: 10.3969/j.issn.1671-4091.2026.04.003 Abstract ( 90 )   ABSTRACT ： Objective ： To investigate the relationship between serum Reverse Triiodothyronine（rT3）and sarcopenia in the patients undergoing maintenance hemodialysis （MHD）Methods：The clinical data of MHD were collected. Pre-dialysis serum biochemical parameters and thyroid function indicators were systematically obtained and recorded. Body composition was assessed via bioelectrical impedance spectroscopy. Functional metrics included grip strength, 6-meter gait speed. Binary logistic regression identified clinical parameters associated with sarcopenia in MHD Results：Sarcopenia patients showed lower serum rT3 levels than non-sarcopenia controls (P=0.022). Both univariate (P<0.05) and multivariate analyses confirmed an independent inverse association between serum rT3 levels and sarcopenia risk （OR=0.495，95％CI=0.268~0.915，P=0.024）. Conclusion serum rT3 levels in MHD patients are independently linked to sarcopenia, with the underlying mechanism. Metrics

Select

The relationship between different nutritional assessment tools and frailty and adverse outcomes in elderly patients undergoing maintenance hemodialysis GAO Jin-xin, RONG Peng, LI Wen-jing, JIN Fang 2026, 25 (04):  289-294.  doi: 10.3969/j.issn.1671-4091.2026.04.004 Abstract ( 118 )   Objective To investigate the efficacy of different nutritional assessment tools in evaluating the nutritional status of elderly patients undergoing maintenance hemodialysis (MHD) and analyze their correlation with frailty and adverse outcomes. Methods A total of 146 elderly patients (aged ≥65 years) undergoing MHD were enrolled. Nutritional status was assessed using the Nutritional Risk Screening 2002 (NRS2002), Mini-Nutritional Assessment-Short Form (MNA-SF), and Malnutrition Inflammation Score (MIS), followed by consistency analysis. Patients were divided into frailty and non-frailty groups, as well as survival and death groups, based on the FRAIL scale (Fatigue, Resistance, Ambulation, Illness, Loss of Weight) and adverse outcomes. Differences in clinical data and nutritional status between groups were compared. Multivariate logistic regression was used to analyze the relationship between NRS2002, MNA-SF, MIS, and frailty. Kaplan-Meier survival analysis and multivariate Cox regression were employed to evaluate the relationship between NRS2002, MNA-SF, MIS, and prognosis.  Results The incidence of malnutrition assessed by NRS2002, MNA-SF, and MIS was 41.78%, 43.84%, and 52.74%, respectively. MIS showed good consistency with NRS2002 and MNA-SF (Kappa=0.701, 0.714), while NRS2002 and MNA-SF had low consistency (Kappa=0.427). The proportion of malnutrition assessed by NRS2002 (χ²=5.385, P=0.020), MNA-SF (χ²=5.023, P=0.025), and MIS (χ²=8.119, P=0.004) was significantly higher in the frailty group than in the non-frailty group. Multivariate logistic regression revealed that NRS2002 (OR=3.903, 95%CI: 1.393～10.937, P=0.010) and MIS (OR=4.574, 95%CI: 1.435～14.585, P=0.010) were independent risk factors for frailty. The proportion of malnutrition assessed by NRS2002 (χ²=10.905, P＜0.001), MNA-SF (χ²=6.837, P=0.009), and MIS (χ²=12.296, P＜0.001) was significantly higher in the death group than in the survival group. Kaplan-Meier analysis showed that the cumulative survival rate of malnourished patients (assessed by NRS2002, MNA-SF, MIS) was significantly lower than that of patients with normal nutrition (χ²=7.375, 9.681, 12.252, P=0.007, 0.002, <0.001, respectively). Multivariate Cox regression indicated that NRS2002 (HR=2.587, 95%CI: 1.210～5.533, P=0.014), MNA-SF (HR=2.663, 95%CI: 1.218～5.821, P=0.014), and MIS (HR=3.697, 95%CI: 1.494～9.147, P=0.005) were independent risk factors for adverse outcomes.  Conclusion  MIS shows good consistency with NRS2002 and MNA-SF in assessing nutritional status in elderly MHD patients. Both MIS and NRS2002 are risk factors for frailty and adverse outcomes and may be more suitable for clinical prognosis evaluation.. Metrics

Select

Risk investigation and influencing factors of sarcopenia in maintenance hemodialysis patients WU Yang, LEI Sheng-ying, GUO Yu-hong 2026, 25 (04):  295-299.  doi: 10.3969/j.issn.1671-4091.2026.04.005 Abstract ( 101 )   Objective  To explore the risk and influencing factors of sarcopenia in Maintenance Hemodialysis (MHD) patients.  Methods  A cross-sectional study was conducted. MHD patients admitted to Nanyang Central Hospital from June 2023 to January 2025 were observed and divided into a sarcopenia group and a non-sarcopenia group based on whether sarcopenia occurred. Clinical data of the patients were collected, and risk factors for sarcopenia in MHD patients were analyzed using a multivariate model.  Results  A total of 150 patients were enrolled, including 58 patients (38.67%) with sarcopenia. Compared with the non-sarcopenia group (n=92), the sarcopenia group had higher age (t=5.339, P<0.001), longer dialysis vintage (t=7.191, P<0.001), higher C-reactive protein (CRP) levels (t=13.284, P<0.001), lower grip strength (t=8.791, P<0.001), lower albumin (Alb) levels (t=6.035, P<0.001), and a higher proportion of patients with diabetes (χ²=13.581, P<0.001). Univariate logistic regression analysis showed that age (OR=1.031, 95%CI: 1.009～1.053, P<0.001), dialysis vintage (OR=1.422, 95%CI: 1.308～1.546, P<0.001), diabetes (OR=3.219, 95%CI: 1.756～5.896, P<0.001), grip strength (OR=0.921, 95%CI: 0.904～0.939, P<0.001), CRP (OR=1.795, 95%CI: 1.662～1.940, P<0.001), and Alb (OR=0.874, 95%CI: 0.836～0.914, P<0.001) were risk factors for sarcopenia in MHD patients. Multivariate logistic regression analysis revealed that age (OR=1.368, 95%CI: 1.102～1.655, P=0.001), dialysis vintage (OR=1.462, 95%CI: 1.112～1.800, P<0.001), diabetes (OR=1.531, 95%CI: 1.118～1.905, P<0.001), grip strength (OR=0.846, 95%CI: 0.747～0.958, P=0.008), CRP (OR=1.468, 95%CI: 1.105～1.785, P<0.001), and Alb (OR=0.621, 95%CI: 0.475～0.812, P<0.001) were independent risk factors for sarcopenia in MHD patients.   Conclusion  Age, dialysis vintage, diabetes, grip strength, CRP, and Alb are independent risk factors for sarcopenia in MHD patients and can be used for early assessment of the risk of sarcopenia in these patients. Metrics

Select

Analysis on the current status and influencing factors of unplanned disconnection during extracorporeal membrane oxygenation combined with continuous renal replacement therapy in patients after lung transplantation FENG Shi-ping, WANG Hai-xiang, QIAN Xiu-qun, HUANG Hao 2026, 25 (04):  300-304.  doi: 10.3969/j.issn.1671-4091.2026.04.006 Abstract ( 60 )   Objective  To explore the current status and influencing factors of unplanned disconnection from continuous renal replacement therapy(CRRT) in patients after lung transplantation during treatment with extracorporeal membrane oxygenation(ECMO) combined with CRRT.  Methods  A retrospective analysis was conducted on the clinical data of 67 patients who underwent lung transplantation and received ECMO combined with CRRT treatment at Wuxi People’s Hospital from February 2018 to March 2024. Based on the criteria for unplanned CRRT disconnection, the patients were divided into the planned CRRT disconnection group and the unplanned CRRT disconnection group. After screening for potential influencing factors using univariate analysis, multivariate Logistic regression analysis was further performed to investigate the independent influencing factors of unplanned CRRT disconnection in post-lung transplant patients during ECMO combined with CRRT treatment. Results  A total of 67 post-lung transplant patients were included in this study, with a total of 211 episodes of ECMO combined with CRRT treatment. During the period, 119 episodes of unplanned CRRT disconnection occurred, with an incidence rate of 56.4%. Multivariate Logistic regression analysis showed that ECMO mode (OR=0.089, 95%CI:0.027～0.301, P＜0.001), average ECMO blood flow rate (OR=18.779, 95% CI:6.754～52.211, P＜0.001), and average CRRT ultrafiltration rate (OR=1.024, 95%CI:1.013～1.034, P＜0.001) were independent risk factors for unplanned CRRT disconnection in patients undergoing combined ECMO and CRRT therapy after lung transplantation. Conclusion During ECMO combined with CRRT therapy in patients after lung transplantation, veno-arterial ECMO (VA-ECMO) support, high ECMO blood flow rate, and excessively high CRRT ultrafiltration rate are all independent risk factors for unplanned CRRT disconnection. These findings suggest that clinical medical staff should develop refined treatment plans and proactively address equipment alarms and pipeline-related issues to reduce the incidence of unplanned CRRT disconnection, thereby improving the prognosis of patients after lung transplantation. Metrics

Select

A cross-sectional survey on medication adherence and social support among maintenance hemodialysis patients YU Hai-yan, WANG Xin-xin, CHEN Di, ZHOU Ran, SHI Ying, CHENG Jun, MA Ying-chun 2026, 25 (04):  305-309.  doi: 10.3969/j.issn.1671-4091.2026.04.007 Abstract ( 83 )   Objective  To investigate medication adherence and social support status among patients undergoing maintenance hemodialysis (MHD).  Methods  Patients receiving MHD at Beijing Boai Hospital of the China Rehabilitation Research Center from January 1 to April 15, 2025, were selected as study subjects. Demographic and medical data were collected via the Renal Rehabilitation Electronic Information Management Platform. Medication adherence was assessed using the Morisky Adherence Scale, while social support was evaluated using the Social Support Rating Scale. Anxiety and depression states were assessed with the Self-Rating Anxiety Scale and Self-Rating Depression Scale, respectively. A logistic regression model was applied to analyze factors influencing medication adherence.  Results  A total of 99 patients were included, with 45.45% in the good adherence group and 54.55% in the poor adherence group. Compared with the poor adherence group, patients in the good adherence group exhibited lower serum potassium (t=2.015, P=0.047) and phosphorus (t=2.621, P=0.010) levels, higher urea clearance index (t=-2.769, P=0.007), and reduced anxiety scores (t=2.489, P=0.015). No statistically significant differences were observed in social support scores (t=1.160, P=0.249), subjective support scores (t=0.687, P=0.494), objective support scores (t=1.450, P=0.151), or support utilization (t=0.410, P=0.682) between the two groups. No significant correlation was found between medication adherence and social support scores (β=-0.139, P=0.169). Patients with higher satisfaction with social support exhibited relatively poorer medication adherence (χ²=5.727, P=0.014). Logistic regression analysis revealed that higher anxiety levels (OR=0.933, 95% CI：0.878～0.992, P=0.025), lower urea clearance index (OR=10.887, 95% CI 1.813～65.364, P=0.009), and greater satisfaction with social support (OR=0.362, 95% CI：0.138～0.951, P=0.039) were associated with poorer medication adherence.  Conclusion  Poor medication adherence is prevalent among MHD patients, and there is a complex interplay between medication adherence and social support. Nursing staff should provide health education to patients through various methods to enhance their social support and medication adherence. Metrics

Select

Frailty mediates the causal relationship between chronic kidney disease and appendicular lean mass: a Mendelian randomization study MA Chen-hong, ZHANG Yu-cai, YAO Shu-ge, LI Cui-xia, ZHANG Li-jun, GUO Yun-ling 2026, 25 (04):  310-315.  doi: 10.3969/j.issn.1671-4091.2026.04.008 Abstract ( 70 )   Objective To investigate frailty index (FI), edema, and depression in the genetic causal relationship between chronic kidney disease (CKD) and appendicular lean mass (ALM), in order to provide references for the prevention of CKD-related sarcopenia.  Methods  Summary data for FI, edema, depression, CKD, and ALM were obtained from publicly available genome-wide association study (GWAS) databases. A two-step Mendelian randomization (MR) approach was first applied to evaluate potential mediators between CKD and ALM. Subsequently, multivariable Mendelian randomization (MVMR) analysis was conducted to identify causal factors independently associated with ALM. Finally, a mediation MR model was constructed, with CKD as the exposure, ALM as the outcome, and the identified independent causal factors as mediators to explore potential mediation pathways.  Results  Two-step MR, MVMR, and mediation MR analyses revealed that FI is a potential independent mediator between CKD and ALM (OR=0.458, 95% CI: 0.411～0.511, P=0.021). Further mediation analysis indicated a significant indirect causal effect of CKD on ALM through FI (βmediation effect=-0.24, OR=0.787, 95% CI: 0.67～0.978, P=0.019).  Conclusion  The effect of CKD on ALM is primarily mediated indirectly through frailty. Frailty serves as a key mediator of CKD-related sarcopenia, providing a genetic basis for clinical intervention of sarcopenia. Metrics

Select

A systematic review of physical activity and exercise interventions in peritoneal dialysis patients WANG Jia, WU Tong-xuan, HE Na , TANG Wen 2026, 25 (04):  316-323.  doi: 10.3969/j.issn.1671-4091.2026.04.009 Abstract ( 72 )   Objective  To evaluate the effects of physical activity and exercise training on patients undergoing peritoneal dialysis (PD). Methods  A systematic search was conducted in MEDLINE, Embase, CINAHL, the Cochrane Central Register of Controlled Trials, CNKI, Wanfang Chinese Academic Journal Database, and Sinomed to collect studies on physical activity and exercise training in PD patients published from database inception to January 2024. Quality assessment was performed, and relevant evidence was extracted and summarized.  Results  A total of 25 interventional studies (n=1692) and 16 observational studies (n=1876) were included. The results indicated that physical activity and exercise training interventions could improve physical function, quality of life, nutritional status, and body composition in PD patients, with no serious adverse events reported. However, the effects of physical activity and exercise training on fatigue, bone mineral density, and blood pressure levels in PD patients remained inconsistent.  Conclusion  The findings of this study support the positive role of physical activity and exercise training in improving multiple health outcomes in PD patients. Future research should focus on developing evidence-based individualized exercise prescriptions and emphasize long-term adherence to intervention protocols to maximize clinical benefits. Metrics

Select

Adequacy assessment of continuous renal replacement therapy LIU Chen-xi, ZHAO Juan-yong, FAN Xin-yan, TANG Xun-zi, YUAN Fang, LIU Hong 2026, 25 (04):  325-329.  doi: 10.3969/j.issn.1671-4091.2026.04.010 Abstract ( 131 )   Continuous renal replacement therapy (CRRT) is a extracorporeal blood purification method to continuously and slowly clear water and solute from blood, and has become a critical life-support technology for critically ill patients. With the advances of technology and the accumulation of clinical experience, the assessment of CRRT "adequacy" has expanded beyond traditional issue of solute clearance dosage to include multiple dimensions such as treatment timing, fluid management, anticoagulation strategies, and circuit monitoring. Based on the latest evidence-based medical research, this article reviews key elements of CRRT adequacy assessment, including selection of initial time, CRRT dosage, surveillance of solute clearance, volume management strategies, anticoagulation protocols, management of specific patients, and quality control system, in order to optimize CRRT practice and improve patient outcomes. Metrics

Select

Systematic review and meta-analysis of risk factors for vascular access thrombosis in patients on maintenance hemodialysis QU Jia-yuan, LI Dan-yang, GUO Mei-qian, SHI Zi-yue, DENG Xiao-mei, WANG Pei-xi 2026, 25 (04):  330-334.  doi: 10.3969/j.issn.1671-4091.2026.04.011 Abstract ( 70 )   Objective  To systematically evaluate the risk factors for vascular access thrombosis (VAT) in patients undergoing maintenance hemodialysis, and to provide evidence-based guidance for clinical intervention.  Methods  A comprehensive literature search was conducted in PubMed, Scopus, Ovid, Web of Science, Cochrane Library, CNKI, Wan Fang, VIP, CBM and Duxiu, covering case-control and cohort studies on VAT-related risk factors from database inception to February, 2025. Meta-analysis was performed using Stata version 17.0.  Results  A total of 17 studies were included, all with a quality assessment score of ≥7. Meta-analysis revealed that female (OR=1.481, 95% CI: 1.177～1.862, P＜0.001), graft arteriovenous fistula (OR=2.773, 95% CI: 1.573～4.889, P＜0.001), diabetes mellitus (OR=2.163, 95% CI: 1.451～3.225, P＜0.001), intradialytic hypotension (OR=2.897, 95% CI: 2.146～3.912, P＜0.001), elevated mean platelet volume (OR=1.587, 95% CI: 1.331～1.891, P＜0.001), and increased TNF-α level (OR=1.132, 95% CI: 1.062～1.206, P＜0.001) were associated with an increased risk of VAT. In contrast, higher hemoglobin level (OR=0.867, 95% CI:0.789～0.952, P＜0.050) and antiplatelet therapy (OR=0.669, 95% CI: 0.567～0.790, P＜0.050) were associated with a reduced risk of VAT.  Conclusion   Female, prosthetic graft, diabetes, intradialytic hypotension, increased mean platelet volume, and elevated TNF-α level are the risk factors for VAT, while higher hemoglobin level and antiplatelet therapy are the protective factors for VAT. Given the observed heterogeneity among the included studies, further research is warranted to confirm these findings across different patient groups. Metrics

Select

Correlation of serum levels of MIP-1α, YKL-40 and NLRP3 with catheter-related infections and prognosis ZHANG Wen-qiang, LIU Hong-fei, QIU Yi 2026, 25 (04):  335-340.  doi: 10.3969/j.issn.1671-4091.2026.04.012 Abstract ( 61 )   Objective  To investigate the correlation of serum levels of macrophage inflammatory protein-1α (MIP-1α), chitinase-3-like protein 1 (YKL-40), and NOD like receptor family pyrin domain-containing protein 3 (NLRP3) with catheter-related infections and prognosis in maintenance hemodialysis (MHD) patients.  Methods A total of 296 MHD patients treated in Luzhou Hospital of Traditional Chinese Medicine from January 2020 to August 2024 were included in this study. They were divided into infected group (n=68) and uninfected group (n=228) based on the presence or absence of catheter infection. The catheter infection patients were further divided into good prognosis subgroup (n=42) or poor prognosis subgroup (n=26). ELISA was applied to detect serum levels of MIP-1α, YKL-40, and NLRP3. Clinical data were collected and analyzed. Multivariate logistic regression was applied to analyze the factors influencing prognosis. Receiver operating characteristic (ROC) curves were plotted to analyze the predictive value of MIP-1α, YKL-40, and NLRP3 for prognosis. Pearson and Spearman methods were applied to analyze the correlation of MIP-1α, YKL-40 and NLRP3 with inflammatory factors and prognosis.  Results  Compared with the uninfected group, the infected group had obviously higher levels of serum MIP-1α, YKL-40, NLRP3, C-reactive protein (CRP), tumor necrosis factor-α (TNF-α), and procalcitonin (PCT) (t=14.532, 14.561, 22.823, 28.495, 11.487 and 22.234 respectively; P<0.001). Dialysis duration (t=3.293, P=0.002), dialysis frequency (t=3.293, P=0.001), MIP-1α (t=8.410, P＜0.001), YKL-40 (t=7.432, P＜0.001), NLRP3 (t=7.511, P＜0.001), CRP (t=15.045, P＜0.001), TNF-α (t=3.694, P＜0.001) and PCT (t=4.007, P＜0.001) levels in the poor prognosis subgroup were obviously higher than those in the good prognosis subgroup. ROC curve showed that the AUC of combined use of MIP-1α, YKL-40 and NLRP3 levels for predicting prognosis was obviously higher than that of single use of MIP-1α (Z=2.101, P=0.036), YKL-40 (Z=2.610, P=0.009) or NLRP3 (Z=2.965, P=0.003). MIP-1α, YKL-40, and NLRP3 were positively correlated with the inflammatory factors of CRP (r=0.462, 0.466 and 0.472 respectively; P＜0.001), TNF-α (r=0.450, 0.444 and 0.455 respectively; P＜0.001), and PCT (r=0.521, 0.502 and 0.504 respectively; P＜0.001), and with prognosis (r=0.521, 0.502 and 0.504 respectively; P＜0.001).   Conclusion Serum MIP-1α, YKL-40 and NLRP3 levels correlate with catheter-related infections and prognosis in MHD patients. Metrics

Select

A hemodialysis mode decision support model based on blood pressure and complications: construction and validation of a local large language model#br# LIU Cheng-cheng, ZHOU Hui, HUANG Li-wa, XU Xiao-min, CHEN Huan-huan, LI Hong-shao 2026, 25 (04):  341-346.  doi: 10.3969/j.issn.1671-4091.2026.04.013 Abstract ( 84 )   Objective To develop an intelligent auxiliary decision-making system for hemodialysis mode based on the Qwen2.5～7B-Instruct large language model, and to evaluate its clinical application effects.  Methods Blood pressure standards and 11 dialysis mode determination rules were integrated. A structural input paradigm was established using 33 virtual cases to optimize the prompt template. Data from 192 patients with maintenance hemodialysis (MHD) at Wenzhou People's Hospital, comprising 940 treatment cycles, were used as the training set. The Low-Rank Adaptation (LoRA) technique was employed for parameter fine-tuning and multi-round prompt engineering optimization. The model performance was evaluated using Kappa consistency, accuracy, precision, recall, and F1 score. A prospective validation study was conducted with 232 MHD patients from April to May 2025. They were divided into the model group (n=58) and the control group (n=174). Blood pressure control, incidence of complications, and evaluation efficiency were compared between the two groups.  Results  After 4 batches of training, the Kappa value between the model's decisions and labels improved from 0.610 to 1.000. The consistency with nurse decisions increased from 0.487 to 0.726, which was higher than the inter-nurse consistency (Kappa=0.612). Using the agreement of two nurses as the standard, the model's accuracy for real cases improved from 56.00% to 80.95%, and the weighted F1 score increased from 0.40 to 0.72. Using the agreement of one nurse as the standard, the accuracy and precision reached 100.00% after training. In clinical validation, 96.55% (56/58) of the model's decisions were recognized by nurses. Compared to the control group, the model group showed no significant differences in blood pressure control during dialysis (χ²=5.744, P=0.057), blood pressure control after dialysis (χ²=0.747, P=0.688), headaches during dialysis (P＞0.999), cramps (χ²=0.347, P=0.541), and manifestations of disequilibrium syndrome (P=0.575). However, the assessment time before machine operation was significantly shortened (t=45.300, P＜0.001).  Conclusion   The large language model optimized with real cases can achieve accurate and rapid decision-making for MHD dialysis modes, demonstrating good clinical applicability and potential for wider adoption. Metrics

Select

Construction and validation of a nomogram prediction model for depression in peritoneal dialysis patients ZHANG Xin, XU Peng-fei, LI Xu, YU Wei-wei, WU Yan, WANG Xiao-yan 2026, 25 (04):  347-352.  doi: 10.3969/j.issn.1671-4091.2026.04.014 Abstract ( 81 )   【Abstract】 Objective To explore the risk factors for depression in peritoneal dialysis (Peritoneal Dialysis, PD) patients and construct a nomogram prediction model. Methods A total of 233 PD patients were selected from Zhongda Hospital Southeast University, Nanjing Drum Tower Hospital and Nanjing Lishui People's Hospital between October 2023 and June 2024 to form the modeling set, while 66 PD patients from July to October 2024 were included in the validation set. Binary logistic regression analysis was employed to identify risk factors for depression, and a nomogram prediction model for depression was constructed and validated.. Results In the modeling set, the anxiety scores (OR=27.579, 95%CI: 12.041~63.165, P＜0.001), work status (OR=0.351, 95%CI: 0.147~0.834, P=0.018), cognitive function scores (OR=0.386, 95%CI: 0.165~0.902, P=0.028), physical health composite index (OR=0.940, 95%CI: 0.899~0.983, P=0.007), hypersensitive C-reactive protein level (OR=1.024, 95%CI: 1.002~1.046, P=0.030), and parathyroid hormone (OR=1.001, 95%CI: 1.000~1.003, P=0.022) were predictive factors for depression in PD patients. The Area Under Curve (AUC) of the Receiver Operating Characteristic Curve for the modeling set was 0.893 (95% CI: 0.852~0.934), with a sensitivity of 76.4% and a specificity of 90.2%. The AUC for the validation set was 0.926 (95% CI: 0.866~0.985), with a sensitivity of 96.0% and a specificity of 82.9%. Conclusion The nomogram prediction model for depression in PD patients demonstrates good clinical utility, aiding health care workers in identifying individuals at high risk of depression. Metrics