Liao Jiali,Li Gefang,He Bo,et al.Construction of a machine learning-based prediction model for Staphylococcus aureus infection in children with septic arthritis[J].Journal of Clinical Pediatric Surgery,2026,(05):477-483.[doi:10.3760/cma.j.cn101785-202507035]
基于机器学习算法的儿童化脓性关节炎金黄色葡萄球菌感染预测模型构建
- Title:
- Construction of a machine learning-based prediction model for Staphylococcus aureus infection in children with septic arthritis
- Keywords:
- Staphylococcus Aureus; Arthritis; Infectious; Machine Learning; Surgical Procedures; Operative; Child
- 摘要:
- 目的 基于化脓性关节炎患儿入院临床特征及实验室检查指标,采用4种机器学习算法构建金黄色葡萄球菌感染预测模型,以期在病原学检测结果出具前判断患儿金黄色葡萄球菌感染状态。方法 回顾性收集2014年1月至2024年12月重庆医科大学附属儿童医院收治的463例化脓性关节炎患儿病例资料,其中2014年1月至2023年12月的380例用于模型构建,2024年1月至12月的83例用于外部验证。建模组通过最小绝对收缩与选择算子(least absolute shrinkage and selection operator,LASSO)回归算法筛选变量,纳入人口学资料、入院症状、入院前处理及实验室检查指标共25个变量。采用五折交叉验证对比极端梯度上升(extreme gradient boosting,XGBoost)、随机森林(random forest,RF)、梯度提升决策树(gradient boosting decision tree,GBDT)及决策树(decision tree,DT)4种机器学习算法构建模型的性能,外部数据集中采用受试者工作特征曲线下面积(area under curve,AUC)、特异度、准确率、精确率、灵敏度、F1分数、决策曲线及校准曲线评估各模型性能。结果 RF与XGBoost算法构建的预测模型,在五折交叉验证及外部验证中均表现良好,AUC值均为0.83、特异度0.82、准确率0.76、灵敏度0.65;其中最重要的特征变量为中性粒细胞/淋巴细胞计数比值。结论 4种机器学习算法构建的预测模型中,RF模型与XGBoost模型基于临床表现及实验室检查指标的预测性能优异,具有良好的预测能力,可在病原学培养结果出具前,为儿童化脓性关节炎金黄色葡萄球菌感染的早期诊断提供辅助决策支持。
- Abstract:
- Objective Toconstruct predictive models for Staphylococcus aureus (S.aureus) infection using 4 machine learning algorithms based on clinical characteristics and laboratory indicators at admission in children with septic arthritis,aiming to determine the infection status of S.aureus before the availability of etiological test results.Methods A retrospective analysis was conducted on the clinical data of 463 children with septic arthritis admitted to the Children’s Hospital of Chongqing Medical University between January 2014 and December 2024.Among them,380 cases from January 2014 to December 2023 were used for model construction (internal modeling cohort),and 83 cases from January 2024 to December 2024 were used for external validation.In the modeling cohort,the least absolute shrinkage and selection operator (LASSO) regression algorithm was used for variable selection.A total of 25 variables were included,covering demographic characteristics,admission symptoms,pre-admission management,and laboratory test indicators.Five-fold cross-validation was used to compare the performance of four machine learning algorithms-extreme gradient boosting (XGBoost),random forest (RF),gradient boosting decision tree (GBDT),and decision tree (DT).In the external validation dataset,model performance was evaluated using the area under the receiver operating characteristic curve (AUC),specificity,accuracy,precision,recall,F1 score,decision curve analysis,and calibration curves.Results The predictive models constructed using the RF and XGBoost algorithms showed good performance in both five-fold cross-validation and external validation,with an AUC of 0.83,specificity of 0.82,accuracy of 0.76,and sensitivity of 0.65.The most important predictive feature was the neutrophil-to-lymphocyte count ratio.Conclusion Among the prediction models constructed using four machine learning algorithms,the RF and XGBoost models based on clinical manifestations and laboratory indicators showed superior predictive performance and good predictive capability.These models may provide supportive decision-making for the early diagnosis of S.aureus infection in children with septic arthritis before the availability of pathogen culture results.
参考文献/References:
[1] Cohen E, Katz T, Rahamim E, et al.Septic arthritis in children:Updated epidemiologic, microbiologic, clinical and therapeutic correlations[J].Pediatr Neonatol, 2020, 61(3):325-330.DOI:10.1016/j.pedneo.2020.02.006.
[2] 陈佳元, 王恩波.儿童化脓性髋关节炎的诊治研究进展[J].中国骨与关节杂志, 2022, 11(2):117-121.DOI:10.3969/j.issn.2095-252X.2022.02.008. Chen JY, Wang EB.Recent advances in the diagnosis and treatment of septic arthritis of the hip in children[J].Chin J Bone Joint, 2022, 11(2):117-121.DOI:10.3969/j.issn.2095-252X.2022.02.008.
[3] Mehler K, Oberthür A, Yagdiran A, et al.Impact of a pediatric infectious disease consultation service on timely step-down to oral antibiotic treatment for bone and joint infections[J].Infection, 2023, 51(4):831-838.DOI:10.1007/s15010-022-01934-4.
[4] Miller RP, Berlouis ME, Hall AG, et al.Effects of antibiotics on α-toxin levels during Staphylococcus aureus culture:implications for the protection of chondrocytes in a model of septic arthritis[J].Cartilage, 2021, 12(3):362-376.DOI:10.1177/1947603519828433.
[5] Aguilar-Gómez NE, Merida-Vieyra J, Isunza-Alonso OD, et al.Surveillance of osteoarticular infections caused by Staphylococcus aureus in a paediatric hospital in Mexico City[J].Front Cell Infect Microbiol, 2022, 12:999268.DOI:10.3389/fcimb.2022.999268.
[6] Tarabichi S, Goh GS, Zanna L, et al.Time to positivity of cultures obtained for periprosthetic joint infection[J].J Bone Joint Surg Am, 2023, 105(2):107-112.DOI:10.2106/JBJS.22.00766.
[7] 黎梓宇, 李各芳, 何波, 等.儿童化脓性关节炎病原菌分布及耐药性分析[J].重庆医科大学学报, 2024, 49(3):290-294.DOI:10.13406/j.cnki.cyxb.003449. Li ZY, Li GF, He B, et al.Distribution and drug resistance of pathogenic bacteria in children with septic arthritis[J].J Chongqing Med Univ, 2024, 49(3):290-294.DOI:10.13406/j.cnki.cyxb.003449.
[8] Lu HH, Uddin S, Hajati F, et al.A patient network-based machine learning model for disease prediction:the case of type 2 diabetes mellitus[J].Appl Intell, 2022, 52(3):2411-2422.DOI:10.1007/s10489-021-02533-w.
[9] Garbo V, Venuti L, Boncori G, et al.Severe Panton-valentine-leukocidin-positive Staphylococcus aureus infections in pediatric age:a case report and a literature review[J].Antibiotics (Basel), 2024, 13(12):1192.DOI:10.3390/antibiotics13121192.
[10] Chang J, Kerr D 4th, Zheng M, et al.Chondrocyte invasion may be a mechanism for persistent Staphylococcus aureus infection in vitro[J].Clin Orthop Relat Res, 2024, 482(10):1839-1847.DOI:10.1097/CORR.0000000000003074.
[11] Wang XM, Guo ZY, Zhang X, et al.Inappropriate empirical antibiotic therapy was an independent risk factor of pediatric persistent S.aureus bloodstream infection[J].Eur J Pediatr, 2023, 182(2):719-729.DOI:10.1007/s00431-022-04729-9.
[12] 李媛媛, 李沁原, 张光莉, 等.儿童金黄色葡萄球菌脓毒症抗生素治疗现状分析[J].中国当代儿科杂志, 2019, 21(4):387-392.DOI:10.7499/j.issn.1008-8830.2019.04.016. Li YY, Li QY, Zhang GL, et al.Current status of antibiotic therapy for Staphylococcus aureus sepsis in children[J].Chin J Contemp Pediatr, 2019, 21(4):387-392.DOI:10.7499/j.issn.1008-8830.2019.04.016.
[13] Benito N, Martínez-Pastor JC, Lora-Tamayo J, et al.Executive summary:guidelines for the diagnosis and treatment of septic arthritis in adults and children, developed by the GEIO (SEIMC), SEIP and SECOT[J].Enferm Infecc Microbiol Clin (Engl Ed), 2024, 42(4):208-214.DOI:10.1016/j.eimce.2023.07.007.
[14] Ley K, Hoffman HM, Kubes P, et al.Neutrophils:new insights and open questions[J].Sci Immunol, 2018, 3(30):eaat4579.DOI:10.1126/sciimmunol.aat4579.
[15] Salimi M, Karam JA, Willman M, et al.Neutrophil to lymphocyte ratio and periprosthetic joint infection:a systematic review and meta-analysis[J].J Arthroplasty, 2024, 39(3):831-838.DOI:10.1016/j.arth.2023.08.067.
[16] de Jager CPC, van Wijk PTL, Mathoera RB, et al.Lymphocytopenia and neutrophil-lymphocyte count ratio predict bacteremia better than conventional infection markers in an emergency care unit[J].Crit Care, 2010, 14(5):R192.DOI:10.1186/cc9309.
备注/Memo
收稿日期:2025-7-16。
基金项目:重庆市科卫联合疾控科研项目(2026KWJK1008)
通讯作者:李各芳,Email:642772059@qq.com