| 44 | 0 | 122 |
| 下载次数 | 被引频次 | 阅读次数 |
目的 基于机器学习技术构建老年慢性肾脏病患者万古霉素稳态谷浓度的预测模型，并利用SHapley可加性解释(SHAP)方法阐明模型中特征的重要性。方法 回顾性分析2013年11月至2023年5月在医院接受万古霉素治疗并进行治疗药物监测(TDM)的革兰阳性菌感染的老年慢性肾脏病患者。采用XGBoost算法构建万古霉素稳态谷浓度预测模型，并利用SHAP值量化各特征对模型预测的贡献度(特征重要性)及其作用方向(归因分析)，评估模型的应用价值。结果 最终模型的预测性能指标：模型拟合度(R2)为0.530，平均绝对误差(MAE)为4.248 mg·L-1，均方误差(MSE)为28.822(mg·L-1)2。模型预测的绝对准确度和相对准确度分别为67.27%和60.91%。SHAP分析结果显示，万古霉素单次给药剂量和肌酐水平是影响谷浓度的最重要因素。其他影响因素包括给药后采血时间、丙氨酸氨基转移酶、血小板计数、合并用药情况、尿素氮、胱抑素C、总蛋白、血红蛋白和身体质量指数。结论 基于机器学习技术的万古霉素稳态谷浓度预测模型及SHAP分析，可为老年慢性肾脏病患者万古霉素个体化用药提供高效、经济的辅助工具，有助于临床医师优化给药方案，提升治疗效益与用药安全性。
Abstract：AIM To construct a predictive model for vancomycin steady-state trough concentrations in elderly patients with chronic kidney disease using machine learning techniques, and elucidate the importance of features in the model using SHapley Additive exPlanations(SHAP). METHODS A retrospective analysis was conducted on elderly patients with chronic kidney disease and Gram-positive bacterial infection who received vancomycin therapy and underwent therapeutic drug monitoring(TDM) from November 2013 to May 2023. An XGBoost-based predictive model for vancomycin steady-state trough concentrations was developed. SHAP values were subsequently employed to quantify feature importance(contribution magnitude to predictions) and directional impact(attribution analysis) of each variable, followed by evaluation of the model's clinical applicability. RESULTS The final model demonstrated the following performance metrics: model fit(R2) = 0.530, mean absolute error(MAE) = 4.248 mg·L-1, and mean squared error(MSE) = 28.822(mg·L-1)2. The absolute and relative prediction accuracies were 67.27% and 60.91%, respectively. SHAP analysis identified vancomycin single-dose and creatinine level as the most critical factors influencing trough concentrations. Other significant contributors included time of blood collection after drug administration, alanine aminotransferase, platelet count, concomitant medications, urea nitrogen, cystatin C, total protein, hemoglobin, and body mass index. CONCLUSION The machine learning-based predictive model for vancomycin steady-state trough concentration, along with SHAP analysis, provides an efficient and cost-effective adjunctive tool for individualised vancomycin dosing in elderly patients with chronic kidney disease. It assists clinicians in optimising dosing regimens, thereby improving therapeutic outcomes and medication safety.
[1]WANG W H,DAI L L,MA J W,et al.Malnutrition accelerates the occurrence of infectious complications in patients with chronic kidney disease:a cross-sectional survey of 682 patients with chronic kidney disease[J].Nutr Clin Pract,2023,38(5):1167.
[2]BELLO A K,OKPECHI I G,LEVIN A,et al.An update on the global disparities in kidney disease burden and care across world countries and regions[J].Lancet Glob Health,2024,12(3):e382.
[3]李燕,王学彬,陈礼治,等.万古霉素血药浓度与临床疗效及急性肾损伤的关系[J].第二军医大学学报,2021(3):281.
[4]JAMES M T,BHATT M,PANNU N,et al.Long-term outcomes of acute kidney injury and strategies for improved care[J].Nat Rev Nephrol,2020,16(4):193.
[5]中华医学会老年医学分会,中国医师协会老年医学科医师分会,《中华老年医学杂志》编辑委员会.中国老年危重患者营养支持治疗指南(2023)[J].中华老年医学杂志,2023,42(9):1009.
[6]ANGEHRN Z,HALDNA L,ZANDVLIET A S,et al.Artificial intelligence and machine learning applied at the point of care[J].Front Pharmacol,2020,11:759.
[7]CHEN T,HE TQ,BENESTY M,et al.Xgboost:extreme gradient boosting[J].R package version 0.4-2,2015,1(4):1.
[8]WOILLARD J B,LABRIFFE M,DEBORD J,et al.Tacrolimus exposure prediction using machine learning[J].Clin Pharmacol Ther,2021,110(2):361.
[9]WOILLARD J B,LABRIFFE M,DEBORD J,et al.Mycophenolic acid exposure prediction using machine learning[J].Clin Pharma And Therapeutics,2021,110(2):370.
[10]HUANG X H,YU Z,WEI X,et al.Prediction of vancomycin dose on high-dimensional data using machine learning techniques[J].Expert Rev Clin Pharmacol,2021,14(6):761.
[11]LEE S,SONG M,HAN J,et al.Application of machine learning classification to improve the performance of vancomycin therapeutic drug monitoring[J].Pharmaceutics,2022,14(5):1023.
[12]MA P,LIU R X,GU W R,et al.Construction and interpretation of prediction model of teicoplanin trough concentration via machine learning[J].Front Med,2022,9:808969.
[13]HUANG X H,YU Z,BU S H,et al.An ensemble model for prediction of vancomycin trough concentrations in pediatric patients[J].Drug Des Devel Ther,2021,15:1549.
[14]YU Z,JI H H,XIAO J W,et al.Predicting adverse drug events in Chinese pediatric inpatients with the associated risk factors:a machine learning study[J].Front Pharmacol,2021,12:659099.
[15]DUCHARME M P,SLAUGHTER R L,EDWARDS D J.Vancomycin pharmacokinetics in a patient population:effect of age,gender,and body weight[J].Ther Drug Mo nit,1994,16(5):513.
[16]MIZOKAMI F,SHIBASAKI M,YOSHIZUE Y,et al.Pharmacodynamics of vancomycin in elderly patients aged 75years or older with methicillin-resistant Staphylococcus aureus hospital-acquired pneumonia[J].Clin Interv Aging,2013,8:1015.
[17]YAHAV D,ABBAS M,NASSAR L,et al.Attention to age:similar dosing regimens lead to different vancomycin levels among older and younger patients[J].Age Ageing,2019,49(1):26.
[18]李深,唐霁松,曹颖,等.XGBoost模型分析院内心脏骤停患者恢复自主循环的特征[J].临床急诊杂志,2024,25(11):579.
[19]张婷,陈瑀珩,丁兰萍,等.结合群体药动学模型和XGBoost算法预测万古霉素AUC24h[J].中国医院药学杂志,2025,45(8):859.
[20]AHMED K A,IBRAHIM A,GONZALEZ D,et al.Population pharmacokinetics and model-based dose optimization of vancomycin in sudanese adult patients with renal impairment[J].Drug Des Dev Ther,2024,18:81.
[21]PONTICELLI C,SALA G,GLASSOCK R J.Drug management in the elderly adult with chronic kidney disease:a review for the primary care physician[J].Mayo Clin Proc,2015,90(5):633.
[22]万古霉素临床应用剂量专家组.万古霉素临床应用剂量中国专家共识[J].中华传染病杂志,2012,30(11):641.
[23]赵思璇,苏珊,何娜,等.肾功能不全患者万古霉素给药方案的系统评价[J].中国临床药理学杂志,2020,36(5):573.
[24]刘亭亭,赵宇蕾,陈晓燚,等.68例患者万古霉素稳态谷浓度和血药浓度曲线下面积的监测结果分析[J].中国现代应用药学,2024,41(19):2711.
[25]周吉欽,朱从磊,官宁宁,等.万古霉素给药方案优化的研究进展[J].河南大学学报(医学版),2022,41(3):162.
[26]何素稳.血清胱抑素C、血肌酐、尿素在慢性肾脏病诊断中的应用价值分析[J].中国社区医师,2025,41(13):46.
[27]汪琳,杨春艳,许慧芳,等.万古霉素血药谷浓度影响因素及多元线性回归模型的构建[J].内蒙古医科大学学报,2024,46(3):293.
[28]LIU T T,PANG H M,JING L,et al.A population pharmacokinetic model of vancomycin for dose individualization based on serum cystatin C as a marker of renal function[J].J Pharm Pharmacol,2019,71(6):945.
[29]LING J,YANG X P,DONG L L,et al.Utility of cystatin C and serum creatinine-based glomerular filtration rate equations in predicting vancomycin clearance:a population pharmacokinetics analysis in elderly Chinese patients[J].Biopharm Drug Dispos,2024,45(1):58.
[30]盛长城,沈祥春,唐才林,等.基于胱抑素C评估肾功能的万古霉素群体药动学研究[J].中国药学杂志,2018,53(21):1847.
[31]HE N,SU S,YE Z K,et al.Evidence-based guideline for therapeutic drug monitoring of vancomycin:2020 update by the division of therapeutic drug monitoring,Chinese pharmacological society[J].Clin Infect Dis,2020,71(S4):S363.
[32]师少军,张玉.肾脏转运体和/或代谢酶介导药物排泄及药物相互作用的研究进展[J].中国医院药学杂志,2022,42(15):1600.
[33]VAN DER PLOEG T, AUSTIN P C,STEYERBERG E W.Modern modelling techniques are data hungry:a simulation study for predicting dichotomous endpoints[J].BMC Med Res Methodol,2014,14(1):137.
基本信息:
DOI：10.19577/j.1007-4406.2026.08.003
中图分类号:R969
引用信息:
[1]张馨,谢林利,马欢,等.基于XGBoost构建老年慢性肾脏病患者万古霉素血药浓度预测模型[J].中国临床药学杂志,2026,35(08):684-691.DOI:10.19577/j.1007-4406.2026.08.003.
基金信息:
国家卫生健康委医院管理研究所项目（编号NIHAYSZX2556）; 重庆市临床药学重点专科建设项目（渝卫办发[2021]52号）
2026-08-25
2026-08-25