201 | 0 | 126 |
下载次数 | 被引频次 | 阅读次数 |
近年来,发热伴血小板减少综合征(Severe fever with thrombocytopenia syndrome, SFTS)传播范围不断扩大,给公众健康构成严重威胁。目前,尚无特异性治疗药物或已上市疫苗,主要靠对症支持治疗。早期识别重症患者并实施有针对性的干预,往往是降低病死率的关键。我们将91例SFTS患者的急性期血清抗体、细胞因子和相关实验室指标联合分析,通过机器学习的方法,构建出多因素Logistic回归、随机森林、支持向量机和XGBoost四种早期患者轻重症分类模型。在变量筛选阶段,我们依次使用单因素Logistic回归,多重共线性诊断和LASSO回归筛选变量,得到IL-10、IFN-γ、IFN-α、IL-8、AST、IgM OD450、TP和LY%共8个建模变量。结果显示,在早期轻重症分类中表现最佳的是随机森林(AUC=0.889),接下来是多因素Logistic回归(AUC=0.822)以及XGBoost(AUC=0.811),而表现相对较差的是支持向量机(AUC=0.744)。此外,我们借助多因素Logistic回归识别出IFN-γ、IL-8、IgM OD450和LY%四种SFTS重症化独立预测因子。随机森林和XGBoost的变量重要性分析均表明,IL-10可能是最重要的预测因子,且IFN-γ、IL-8和AST也有较高的重要性。本研究基于血清学指标与机器学习方法开展SFTS患者轻重症分类模型的研究,为SFTS早期风险评估提供了新思路。
Abstract:In recent years, the geographical spread of severe fever with thrombocytopenia syndrome(SFTS)has continued to expand, posing a significant threat to public health. Currently, there are no specific antiviraltherapies or approved vaccines available, and management primarily relies on symptomatic and supportive care.Early identification of patients at risk of developing severe disease and timely targeted interventions are critical for reducing mortality. In this study, acute-phase serum antibodies, cytokines, and laboratory parameters from91 patients with SFTS were analyzed. Using machine learning approaches, four early-stage classificationmodels for distinguishing mild and severe cases were constructed, including multivariate logistic regression,random forest, support vector machine(SVM), and XGBoost models. During variable selection, univariatelogistic regression, multicollinearity diagnostics, and LASSO regression were sequentially applied, ultimatelyidentifying eight variables for model development: IL-10, IFN-γ, IFN-α, IL-8, AST, IgM OD450, TP, andLY%. Among the models, the random forest exhibited the best performance(AUC = 0.889), followed bymultivariate logistic regression(AUC = 0.822) and XGBoost(AUC = 0.811), whereas the support vectormachine demonstrated relatively lower performance(AUC = 0.744). In addition, multivariate logisticregression analysis identified IFN-γ, IL-8, IgM OD450, and LY% as independent predictors of progression tosevere SFTS. Variable importance analyses from both the random forest and XGBoost models indicated that IL-10 was likely the most critical predictive factor, with IFN-γ, IL-8, and AST also demonstrating highimportance. This study establishes an early-stage classification model for SFTS severity based on serological indicators combined with machine learning methods, offering new perspectives for early risk stratification and clinical management of SFTS.
[1] Yu XJ, Liang MF, Zhang SY, et al. Fever with thrombocytopenia associated with a novel bunyavirus in China[J]. N Engl J Med, 2011, 364(16):1523-1532.DOI:10. 1056/NEJMoa1010095.
[2] Li DX. Severe fever with thrombocytopenia syndrome:a newly discovered emerging infectious disease[J]. Clin Microbiol Infect, 2015, 21(7):614-620. DOI:10. 1016/j. cmi. 2015. 03. 001.
[3] Li JC, Zhao J, Li H, et al. Epidemiology, clinical characteristics, and treatment of severe fever with thrombocytopenia syndrome[J]. Infect Med(Beijing),2022, 1(1):40-49. DOI:10. 1016/j.imj. 2021. 10. 001.
[4] Wen Y, Song D, Li A, et al. Longitudinal analysis of antibody dynamics in severe fever with thrombocytopenia syndrome patients-high-incidence regions of China, 2010-2023[J]. China CDC Wkly,2024, 6(46):1211-1217. DOI:10. 46234/ccdcw2024. 244.
[5]牛文静,吴泓晓,陈志海.大别班达病毒的传播途径研究进展[J].病毒学报,2025,41(02):600-607. DOI:10. 13242/j. cnki. bingduxuebao. 004667.
[6] Ding YP, Liang MF, Ye JB, et al. Prognostic value of clinical and immunological markers in acute phase of SFTS virus infection[J]. Clin Microbiol Infect, 2014,20(11):O870-O878. DOI:10. 1111/1469-0691. 12636.
[7] Zhu J, Zhou J, Tao C, et al. Identification of early prognostic biomarkers in Severe Fever with Thrombocytopenia Syndrome using machine learning algorithms[J]. Ann Med, 2025, 57(1):2451184.DOI:10. 1080/07853890. 2025. 2451184.
[8]王雪琪,黄龙,张锦华,等.基于多种机器学习算法鉴定四个靶基因作为儿童急性轮状病毒感染的分子生物标志物的研究[J].病毒学报,2025, 41(02):446-461.DOI:10. 13242/j. cnki. bingduxuebao. 250043.
[9]任碧燕,刘璐,舒坤贤,等.人工智能在宿主与病原体蛋白质互作预测中的应用进展[J].病毒学报,2024,40(5):1121-1136. DOI:10. 13242/j. cnki.bingduxuebao. 004561.
[10]黄泰,于琦,穆俊芳,等.新型冠状病毒感染与2型糖尿病共病联系的生物信息学分析[J].病毒学报,2024, 40(01):65-79. DOI:10. 13242/j. cnki.bingduxuebao. 004444
[11]He Q, You Z, Dong Q, et al. Machine learning for identifying risk of death in patients with severe fever with thrombocytopenia syndrome[J]. Front Microbiol,2024, 15:1458670. DOI:10. 3389/fmicb. 2024. 1458670.
[12]Pettit RW, Fullem R, Cheng C, et al. Artificial intelligence, machine learning, and deep learning for clinical outcome prediction[J]. Emerg Top Life Sci,2021, 5(6):729-745. DOI:10. 1042/ETLS20210246.
[13]An Q, Rahman S, Zhou J, et al. A comprehensive review on machine learning in healthcare industry:classification, restrictions, opportunities and challenges[J]. Sensors(Basel), 2023, 23(9):4178. DOI:10. 3390/s23094178.
[14]Al Meslamani AZ, Sobrino I, de la Fuente J. Machine learning in infectious diseases:potential applications and limitations[J]. Ann Med, 2024, 56(1):2362869.DOI:10. 1080/07853890. 2024. 2362869.
[15]Deo RC. Machine learning in medicine[J]. Circulation,2015, 132(20):1920-1930. DOI:10. 1161/circulationaha. 115. 001593.
[16]Ting Sim JZ, Fong QW, Huang W, et al. Machine learning in medicine:what clinicians should know[J].Singapore Med J, 2023, 64(2):91-97. DOI:10. 11622/smedj. 2021054.
[17]Maltarollo VG, Kronenberger T, Espinoza GZ, et al.Advances with support vector machines for novel drug discovery[J]. Expert Opin Drug Discov, 2019, 14(1):23-33. DOI:10. 1080/17460441. 2019. 1549033.
[18]Lin H, Satapathy SC, Rajinikanth V. Medical data assessment with traditional, machine-learning and deeplearning techniques[J]. Curr Med Imaging, 2020,16(10):1185-1186. DOI:10. 2174/157340561610210112143516.
[19]Rajula HSR, Verlato G, Manchia M, et al.Comparison of conventional statistical methods with machine learning in medicine:diagnosis, drug development, and treatment[J]. Medicina(Kaunas),2020, 56(9):455. DOI:10. 3390/medicina56090455.
[20]Qian FH, Cao Y, Liu YX, et al. A predictive model to explore risk factors for severe COVID-19[J]. Sci Rep,2024, 14(1):18197. DOI:10. 1038/s41598-024-68946-y.
[21]Kang SY, Yoo JR, Park Y, et al. Fatal outcome of severe fever with thrombocytopenia syndrome(SFTS)and severe and critical COVID-19 is associated with the hyperproduction of IL-10 and IL-6 and the low production of TGF-β[J]. J Med Virol, 2023, 95(7):e28894. DOI:10. 1002/jmv. 28894.
[22]Liu Z, Jiang Z, Zhang L, et al. A model based on metaanalysis to evaluate poor prognosis of patients with severe fever with thrombocytopenia syndrome[J]. Front Microbiol, 2024, 14:1307960. DOI:10. 3389/fmicb. 2023. 1307960.
[23]Xing X, Guan X, Liu L, et al. A case-control study of risk sources for severe fever with thrombocytopenia syndrome in Hubei Province, China[J]. Int J Infect Dis, 2017, 55:86-91. DOI:10. 1016/j.ijid. 2017. 01. 003.
[24]Hou H, Zou S, Wei W, et al. Kinetics and prognostic significance of laboratory markers in patients with severe fever with thrombocytopenia syndrome:insight from a comprehensive analysis[J]. J Infect Dis, 2024, 229(6):1845-1855. DOI:10. 1093/infdis/jiad426.
[25]Wang B, He Z, Yi Z, et al. Application of a decision tree model in the early identification of severe patients with severe fever with thrombocytopenia syndrome[J].PLoS One, 2021, 16(7):e0255033. DOI:10. 1371/journal. pone. 0255033.
[26]CouronnéR, Probst P, Boulesteix AL. Random forest versus logistic regression:a large-scale benchmark experiment[J]. BMC Bioinformatics, 2018, 19(1):270. DOI:10. 1186/s12859-018-2264-5.
[27]Zhu W, Shen S, Zhang Z. Improved multiclassification of schizophrenia based on XGBoost and information fusion for small datasets[J]. Comput Math Methods Med, 2022, 2022:1581958. DOI:10. 1155/2022/1581958.
[28]Lu S, Xu L, Liang B, et al. Liver function derangement in patients with severe fever and thrombocytopenia syndrome[J]. J Clin Transl Hepatol,2022, 10(5):825-834. DOI:10. 14218/JCTH. 2021. 00345.
基本信息:
DOI:10.13242/j.cnki.bingduxuebao.250108
中图分类号:R512.8
引用信息:
[1]闻闫瀚,田婷婷,黄晓霞等.基于SFTS患者血清学指标与机器学习的轻重症分类模型研究[J].病毒学报,2025,41(03):702-712.DOI:10.13242/j.cnki.bingduxuebao.250108.
基金信息:
国家重点研发计划任务(项目号:2022YFC2303402),题目:病原宿主互作机制研究及广谱新靶点发现~~