切换至 "中华医学电子期刊资源库"

中华临床实验室管理电子杂志 ›› 2026, Vol. 14 ›› Issue (03) : 199 -207. doi: 10.3877/cma.j.issn.2095-5820.2026.03.003

论著

人工智能整合临床检验数据预测慢性肾脏病进展
任天瑞1, 杨红柳2, 应斌武1,()   
  1. 1 610041 四川成都,四川大学华西医院科技部
    2 610041 四川成都,四川大学华西医院肾脏内科/华西肾脏病研究所
  • 收稿日期:2026-05-26 出版日期:2026-08-28
  • 通信作者: 应斌武
  • 基金资助:
    国家自然科学基金青年基金(82100717); 四川省自然科学基金项目(2025ZNSFSC1610)

Artificial intelligence integrating laboratory data for predicting chronic kidney disease progression

Tianrui Ren1, Hongliu Yang2, Binwu Ying1,()   

  1. 1 Science & Technology Department, West China Hospital of Sichuan University, Chengdu Sichuan 610041, China
    2 Department of Nephrology and Institute of Kidney Diseases, West China Hospital, Sichuan University, Chengdu Sichuan 610041, China
  • Received:2026-05-26 Published:2026-08-28
  • Corresponding author: Binwu Ying
引用本文:

任天瑞, 杨红柳, 应斌武. 人工智能整合临床检验数据预测慢性肾脏病进展[J/OL]. 中华临床实验室管理电子杂志, 2026, 14(03): 199-207.

Tianrui Ren, Hongliu Yang, Binwu Ying. Artificial intelligence integrating laboratory data for predicting chronic kidney disease progression[J/OL]. Chinese Journal of Clinical Laboratory Management(Electronic Edition), 2026, 14(03): 199-207.

目的

探索基于临床实验室数据构建人工智能(AI)模型以预测慢性肾脏病(CKD)相关结局的研究进展。

方法

根据PRISMA-ScR指南,在PubMed、Web of Science、Scopus、中国知网及万方等数据库中系统地检索利用AI和临床实验室数据预测CKD结局的文献。

结果

最终纳入16项研究,涵盖6个国家的数据,样本量区间为308例~480万例。随机森林与梯度提升类算法是应用最广泛的建模算法。除估算肾小球滤过率与血肌酐外,尿白蛋白/肌酐比、血清白蛋白、血红蛋白、胱抑素C及糖化血红蛋白等辅助指标具有独立的预测价值,涵盖肾功能、机体代谢、营养状态及血液相关检测多个评估维度。各研究最优模型曲线下面积分布于0.689~0.993。

结论

基于临床检验数据的AI预测模型在CKD多类结局预测中展现出良好的性能潜力,多维度检验指标的综合运用显著优于单一指标评估体系。

Objective

To explore the current landscape of artificial intelligence (AI) models built on clinical laboratory data for predicting chronic kidney disease (CKD)-related outcomes.

Methods

Following the PRISMA-ScR guidelines, a systematic search was conducted across PubMed, Web of Science, Scopus, China National Knowledge Infrastructure(CNKI), and Wanfang Databases to identify studies employing AI algorithms with clinical laboratory data to predict CKD outcomes.

Results

16 studies were ultimately included, spanning data from 6 countries, with sample sizes ranging from 308 to 4.8 million participants. Random forest and gradient boosting algorithms are the most widely adopted modeling algorithms. Beyond estimated glomerular filtration rate and serum creatinine, several auxiliary biomarkers demonstrated independent predictive value including urine albumin-to-creatinine ratio, serum albumin, hemoglobin, cystatin C, and glycosylated hemoglobin spanning the domains of renal function, metabolism, nutrition, and hematology. The best-performing models across included studies achieved area under curve values ranging from 0.689 to 0.993.

Conclusion

AI predictive models built on clinical laboratory data demonstrate promising performance across multiple CKD outcome categories, with the integration of multidimensional laboratory biomarkers offering a clear advantage over conventional single-marker assessment frameworks.

图1 文献筛选流程图
表1 纳入文献的特征(n=16)
序号 作者 基本信息 检验数据特征 建模方法与结局预测 性能指标/验证值
年份/年 国家/地区 数据来源/验证方式 样本量/例 临床特征总数 最终纳入的检验指标 临床结局 时间窗 算法类型
1 Yun等[9] 2026 韩国 LINKA database[10]/内部交叉验证 1444 122 24-h urine,hematocrit,ionized level,BUN,eGFR AKI进展为ESRD 7天 CoxBoost(最佳),Elastic-Net Cox,Random survival forest,Cox proportional hazards models C-index=0.811/0.742
2 Yong等[11] 2026 中国 皖南医学院第二附属医院,NHANES数据库 /外部验证 训练集:3114;
验证集:496。
55 HbA1c,TyG,globulin DM进展为DKD L R(最佳),D T,R F,S V M,XGBoost,LGBM,DNN AUC=0.689
3 Ye等[12] 2026 中国 福建中医药大学附属第二医院/内部验证 训练集:
308;
验证集:
52。
24 urea,apolipoprotein A1,albumin,HDL-C,uric acid,phosphorus,calcium,globulin,total protein,triglycerides,total cholesterol,LDL-C 进展为CKD G1-2;
进展为CKD G3a-5
RF,Gradient Boosting(最佳),Extra Trees,LR,ANN AUC=0.972/0.965
4 Wang等[13] 2026 中国 宁夏医科大学总医院/内部验证 16 800 23 serum sodium,BUN S-AKI进展为死亡 XGBoost(最佳),RF,LR,DT AUC=0.879
5 Tangri等[14] 2026 美国 美国商业保险队列/外部验证 4 800 000 7 serum creatinine,eGFR,尿常规 CKD进展(eGFR下降40%或肾脏损伤) 2年 Klinrisk模型 AUC=(0.830-0.870)/(0.800-0.810)
6 Shu等[15] 2026 中国 武汉市中心医院/内部验证 400 64 urine microalbumin,eGFR,urea,UACR,lymphocytes,HCT,TBIL,DBIL,ALB,osteocalcin,PTH,total protein,MCV CKD2-4期进展为血液透析发生 RF(最佳),XGBoost,ANN,Weighted Voting,Soft Voting,SVM,DT,KNN,Naive Bayes AUC=0.993
7 Nguyen等[16] 2026 德国 既往研究[17,18,19]/3项独立前序队列 训练集:2774;验证集:1190 18 C3,CFB-derived peptides eGFR下降40%,ESRD,死亡 2.91年 SVM,Cox model AUC=0.801
8 Nagasu等[20] 2026 日本 J-CKD-DB-Ex/内部验证 10 474 14 eGFR,serum creatinine,plasma albumin,serum sodium,serum potassium,urine protein qualitative,urine protein eGFR斜率/3年变化轨迹 1年,2年,3年 LGBM(最佳),LSTM,LR RMSE=2.950
9 Ma等[21] 2026 英国 UK biobank/内部验证 233 589 20 cystatin C,HbA1c,CRP,urea,eGFR,M-VLDL-CE,histidine,IGF-1 发生CKD或CKD相关死亡 根据随访时间 Cox proportional hazards model C-index=0.800/0.866
10 Wu等[22] 2025 中国 武汉第四医院/内部验证 12 151 47 eGFR,Hb,CK-MB DM合并CKD进展为eGFR下降30% 1年 LR,XGBoost(最佳),RF,SVM,DT,Naïve Bayes,KNN,GLM with Elastic Net Regularization,Generalized Additive Model,Multivariate Adaptive Splines AUC=0.906/0.768
11 Zhu等[23] 2023 美国 芝加哥大学医学中心/内部验证 82 667 1 eGFR CKD Ⅱ/Ⅲ期进展为CKD Ⅳ/Ⅴ 90天,365天 LSTM,RNN(最佳),RF,LGBM,Dynamic Cox proportional hazards model,Static Cox proportional hazards model AUC=0.967/0.964
12 Zou等[24] 2022 中国 四川大学华西医院/内部验证 390 34 cystatin C,sAlb,Hb,24-h urine,total protein,eGFR 进展为ESRD或接受透析 1年,3年,5年 RF(最佳),LR,SVM,GBM AUC=0.900
13 Su等[25] 2022 中国台湾 台湾全民健康保险管理局/外部验证 858 21 urine creatinine,serum creatinine CKD2-5进展为ESRD 3年,5年 LR,RF(最佳),XGBoost,SVM,GNB AUC=0.990
14 Lee等[26] 2022 中国台湾 台北市老兵总医院大数据中心/内部验证 112 628 41 eGFR,albuminuria,UACR 脓毒血症进展为ESRD 中位随访时间3.5年 GBDT(最佳),LR,RF,XGBoost,LGBM AUC=0.879
15 Inaguma等[27] 2022 日本 富士塔卫生大学医院/内部验证 5657 21 Hb、Alb、CRP eGFR下降≥30% 2年 LR,RF(最佳) AUC=0.790
16 王雅婷等[28] 2024 中国 陆军军医大学第二附属医院/内部验证 400 47 白蛋白、补体C3血清Klotho蛋白 CKD进展为全因死亡 2012-2019 单因素Cox回归、Lasso-Cox模型、多因素Cox回归分析 AUC=0.760,C-index=0.755
表2 16项研究主要算法类型应用频次与代表性研究汇总
表3 16项研究按预测结局模型性能汇总
表4 主要算法的优缺点比较
1
GBD 2023 CHRONIC KIDNEY DISEASE COLLABORATORS. Global, regional, and national burden of chronic kidney disease in adults, 1990–2023, and its attributable risk factors: A systematic analysis for the Global Burden of Disease Study 2023[J]. The Lancet, 2025, 406(10518): 2461-2482.
2
沈逸枫, 朱晶, 杨静, 等. 2021慢性肾脏病流行病学合作研究CKD-EPI和欧洲肾脏功能联盟EKFC估算肾小球滤过率公式的临床应用评估[J]. 中华检验医学杂志, 2024, 47(8): 879-887.
3
KIDNEY DISEASE: IMPROVING GLOBAL OUTCOMES (KDIGO) CKD WORK GROUP. KDIGO 2024 clinical practice guideline for the evaluation and management of chronic kidney disease[J]. Kidney international, 2024, 105(4S): S117-S314.
4
LUNDBERG S M, LEE S I. A unified approach to interpreting model predictions[C/OL]. Conference on Neural Information Processing Systems (NIPS), Long Beach, CA, USA, 2017. [2026-05-13].
5
TRICCO A C, LILLIE E, ZARIN W, et al. PRISMA extension for scoping reviews (PRISMA-ScR): Checklist and explanation[J]. Annals of internal medicine, 2018, 169(7): 467-473.
6
ARKSEY H, O'MALLEY L. Scoping studies: Towards a methodological framework[J]. International journal of social research methodology, 2005, 8(1): 19-32.
7
LEVAC D, COLQUHOUN H, O'BRIEN K K. Scoping studies: Advancing the methodology[J]. Implementation science, 2010, 5(1): 69.
8
MCGOWAN J, SAMPSON M, SALZWEDEL D M, et al. PRESS peer review of electronic search strategies: 2015 guideline statement[J]. Journal of clinical epidemiology, 2016, 75: 40-46.
9
YUN D, HONG A, KIM K, et al. Machine learning survival analysis for predicting kidney disease progression in patients with acute kidney injury undergoing continuous kidney replacement therapy: An analysis of the LINKA database[J]. Journal of critical care, 2026, 92:155419.
10
YUN D, HAN S S, LEE J, et al. Study protocol for a consortium linking health medical records, biospecimens, and biosignals in korean patients with acute kidney injury (LINKA cohort)[J]. Kidney research and clinical practice, 2026, 45(2): 274-283.
11
YONG W, PENG D D, YE K, et al. Machine learning model based on routine blood and biochemical parameters for early diagnosis of diabetic kidney disease[J]. Frontiers in endocrinology, 2026, 17: 1720574.
12
YE B, ZHANG X, ZHU W, et al. Interpretable machine learning model based on routine metabolic laboratory indices to identify advanced chronic kidney disease[J]. Frontiers in endocrinology, 2026, 17: 1776419.
13
WANG J Z, ZHANG N, MA R R, et al. An interpretable machine-learning model for predicting in-hospital mortality in patients with sepsis-associated acute kidney injury[J]. Frontiers in medicine, 2026, 13: 1756831.
14
TANGRI N, FERGUSON T W, TENG C C, et al. Validation of the klinrisk machine learning model for CKD progression in a large representative US population[J]. Journal of the American Society of Nephrology, 2026, 37(2): 326-337.
15
SHU P, QIN D, XU F, et al. Development and internal validation of an interpretable machine learning model for predicting dialysis risk in patients with stage 3–4 chronic kidney disease[J]. Frontiers in public health, 2026, 14: 1782951.
16
NGUYEN T M N, KONDYLI M, MISCHAK H, et al. Association of urinary complement peptides with kidney function and progression of kidney disease[J]. International journal of molecular sciences, 2026, 27(4): 1982.
17
HE T, MISCHAK M, CLARK A L, et al. Urinary peptides in heart failure: A link to molecular pathophysiology[J]. European journal of heart failure, 2021, 23(11): 1875-1887.
18
TOFTE N, LINDHARDT M, ADAMOVA K, et al. Early detection of diabetic kidney disease by urinary proteomics and subsequent intervention with spironolactone to delay progression (PRIORITY): A prospective observational study and embedded randomised placebo-controlled trial[J]. The Lancet. Diabetes & endocrinology, 2020, 8(4): 301-312.
19
ZHANG Z, STAESSEN J A, THIJS L, et al. Left ventricular diastolic function in relation to the urinary proteome: A proof-of-concept study in a general population[J]. International journal of cardiology, 2014, 176(1): 158-165.
20
NAGASU H, NAKASHIMA T, IHARA K, et al. Prediction of estimated glomerular filtration rate slope and kidney prognosis of patients with chronic kidney disease[J]. Scientific reports, 2026, 16(1): 8883.
21
MA J, LIU R, FENG X, et al. Explainable machine learning integrating biochemical and metabolomic biomarkers with conventional clinical factors improves chronic kidney disease prediction and risk stratification[J]. BMC nephrology, 2026, 27(1): 137.
22
WU J, GAO Q, TIAN M, et al. Explainable machine learning prediction of 1-year kidney function progression among patients with type 2 diabetes mellitus and chronic kidney disease: A retrospective study[J]. Monthly journal of the Association of Physicians, 2025, 118(9): 647-656.
23
ZHU Y, BI D, SAUNDERS M, et al. Prediction of chronic kidney disease progression using recurrent neural network and electronic health records[J]. Scientific reports, 2023, 13(1): 22091.
24
ZOU Y, ZHAO L, ZHANG J, et al. Development and internal validation of machine learning algorithms for end-stage renal disease risk prediction model of people with type 2 diabetes mellitus and diabetic kidney disease[J]. Renal failure, 2022, 44(1): 562-570.
25
SU C T, CHANG Y P, KU Y T, et al. Machine learning models for the prediction of renal failure in chronic kidney disease: A retrospective cohort study[J]. Diagnostics (Basel, Switzerland), 2022, 12(10): 2454.
26
LEE K H, CHU Y C, TSAI M T, et al. Artificial intelligence for risk prediction of end-stage renal disease in sepsis survivors with chronic kidney disease[J]. Biomedicines, 2022, 10(3): 546.
27
INAGUMA D, HAYASHI H, YANAGIYA R, et al. Development of a machine learning-based prediction model for extremely rapid decline in estimated glomerular filtration rate in patients with chronic kidney disease: A retrospective cohort study using a large data set from a hospital in Japan[J]. BMJ open, 2022, 12(6): e058833.
28
王雅婷, 熊加川, 赵景宏. 基于血清Klotho蛋白的慢性肾脏病患者全因死亡预测机器学习模型的构建与验证[J]. 陆军军医大学学报, 2024, 46(8): 859-867.
29
HEERSPINK H J L, COLLIER W H, CHAUDHARI J, et al. A meta-analysis of albuminuria as a surrogate endpoint for kidney failure[J]. Nature medicine, 2026, 32(1): 281-287.
30
BAO L, ZHOU Q, LIN Y, et al. Cystatin C as a new biomarker in patients with chronic kidney disease a review and meta-analysis[J]. American journal of biochemistry and biotechnology, 2021, 17(1/3): 328-337.
31
XU J, HE Q, WANG M, et al. Handling time-varying treatments in observational studies: A scoping review and recommendations[J]. Journal of evidence-based medicine, 2024, 17(1): 95-105.
[1] 王燕妮. AI在创面管理中的潜力与挑战[J/OL]. 中华损伤与修复杂志(电子版), 2026, 21(04): 320-320.
[2] 高嘉敏, 韩俊毅. 老年胃肠癌患者多模式预康复:异质性挑战与分类管理策略[J/OL]. 中华普通外科学文献(电子版), 2026, 20(03): 192-199.
[3] 孟凡征, 王阳清, 张珅瑜, 刘连新. 人工智能在肝癌微创外科中的发展[J/OL]. 中华普外科手术学杂志(电子版), 2026, 20(04): 307-310.
[4] 王葵, 林宇轩, 沈皓, 沈锋. 人工智能化在肝癌微创外科的实践[J/OL]. 中华普外科手术学杂志(电子版), 2026, 20(04): 311-314.
[5] 王鑫, 阎龙, 戴朝六. 人工智能化与个体化在肝癌微创外科的应用[J/OL]. 中华普外科手术学杂志(电子版), 2026, 20(04): 315-318.
[6] 李佳颖, 刘溪, 陈宇航, 桂程鹏, 习明, 陈炜, 罗俊航. 卷积神经网络在预测肾透明细胞癌全视野数字切片肿瘤坏死中的应用研究[J/OL]. 中华腔镜泌尿外科杂志(电子版), 2026, 20(04): 389-398.
[7] 黄楚曦, 吴卓. 人工智能在膀胱癌影像中的应用进展[J/OL]. 中华腔镜泌尿外科杂志(电子版), 2026, 20(03): 241-247.
[8] 郑瑞, 何天霖. 人工智能引领胃癌外科治疗迈向精准微创新时代[J/OL]. 中华腔镜外科杂志(电子版), 2026, 19(03): 183-186.
[9] 冯源康, 薄智豪, 王勇. 机器人外科的演进与未来[J/OL]. 中华腔镜外科杂志(电子版), 2026, 19(03): 187-192.
[10] 孙雪梅, 赵海丹. 基于随机对照试验的人工智能导诊系统对三甲医院门诊服务质量的影响研究[J/OL]. 中华临床医师杂志(电子版), 2026, 20(03): 222-227.
[11] 胡伟铭, 陈为谦, 林桂涵, 陈炜越, 胡展宁, 马成, 纪建松. 人工智能在冠心病介入治疗及预后评估中的应用[J/OL]. 中华介入放射学电子杂志, 2026, 14(03): 243-249.
[12] 刘诗馨, 宋晓微, 武剑. 临床视角下人工智能赋能脑卒中康复的范式重塑[J/OL]. 中华脑血管病杂志(电子版), 2026, 20(03): 231-239.
[13] 梁怡凡, 牟婧宇, 吴雅婷, 邳靖陶, 陈乐, 武剑. 不同人工智能模型预测脑卒中不良预后诊断效能的荟萃分析[J/OL]. 中华脑血管病杂志(电子版), 2026, 20(03): 308-319.
[14] 张思远, 宋晓微, 王丽君, 范玉华, 武剑. 人工智能赋能脑小血管病的血流动力学评估:研究进展与临床转化[J/OL]. 中华脑血管病杂志(电子版), 2026, 20(03): 327-333.
[15] 徐林, 简讯, 刘道权, 易东, 鄢华. 人工智能在心脑血管内介入诊疗中的应用进展——从单一影像模式到多模态影像融合[J/OL]. 中华脑血管病杂志(电子版), 2026, 20(03): 334-341.
阅读次数
全文


摘要


AI


AI小编
你好!我是《中华医学电子期刊资源库》AI小编,有什么可以帮您的吗?