LASSO-logistic回归模型在高尿酸血症影响因素分析中的应用

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (1541 KB)
输出: BibTeX | EndNote (RIS)

摘要目的探讨LASSO-logistic回归模型在高尿酸血症(HUA)影响因素分析中的应用。方法回顾性收集2021年在武警宁夏总队医院参加健康体检的男性军队人员健康档案,用logistic回归和LASSO-logistic回归两种模型筛选影响HUA的因素,并用赤池信息准则(AIC)和贝叶斯信息准则(BIC)评价模型拟合效果。结果男性军队人员HUA检出率为27.3%,传统logistic回归和LASSO-logistic回归两个模型的结果均显示,HUA的关键影响因素主要有年龄、丙氨酸氨基转移酶、天门冬氨酸氨基转移酶、γ-谷氨酰转肽酶、肌酐水平和脂肪肝,其中γ-谷氨酰转肽酶[传统logistic回归模型和LASSO-logistic模型OR (95%CI)分别为1.03(1.03,1.04)、1.03(1.02,1.04)]与肌酐[两个模型OR (95%CI)均为1.05(1.05,1.06))]是影响HUA最重要的两个指标;LASSO-logistic回归模型的AIC(4221.373)和BIC(4308.966)均小于传统logistic回归模型(AIC为4223.373,BIC为4317.222)。结论基于LASSO法筛选出的变量组成的LASSO-logistic回归模型拟合效果较好,是研究HUA影响因素的可靠选择。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	谢小莲
	杜涛
	乔静

关键词 ： LASSO-logistic回归模型, 高尿酸血症, 影响因素

Abstract：Objective To study the application of LASSO-logistic regression model in the analysis of the influencing factors of hyperuricemia (HUA). Methods The health records of male military personnel who participated in physical examination in Ningxia Autonomous Regional Corps Hospital of Chinese People's Armed Police Force in 2021 were collected. Two kinds of models were established by traditional logistic regression and LASSO-logistic regression to screen the factors affecting hyperuricemia, and the model fitting effect was evaluated by AIC and BIC. Results The detection rate of HUA in male military personnel was 27.3%. The results of the two models showed that the key influencing factors of hyperuricemia were age, ALT, AST, GGT, Crea and fatty liver, among which GGT [the OR values (95%CI) of traditional logistic regression and LASSO-logistic model were 1.03 (1.03, 1.04) and 1.03 (1.02, 1.04), respectively] and creatinine levels [the OR values (95%CI) of two models were 1.05 (1.05, 1.06)] were the two most important factors affecting hyperuricemia. The AIC(4221.373) and BIC (4308.966) of LASSO-logistic regression model were both lower than that of the traditional logistic regression model(AIC is 4223.373,and BIC is 4317.222) (4308.966). Conclusions The logistic regression model composed of variables screened by LASSO method has a better fitting effect, and it is a reliable choice to study the influencing factors of hyperuricemia.

Key words： LASSO-logistic regression model hyperuricemia influencing factor

收稿日期: 2022-10-28

ZTFLH:

R181.22

作者简介: 谢小莲,硕士,医师。

引用本文:

谢小莲, 杜涛, 乔静. LASSO-logistic回归模型在高尿酸血症影响因素分析中的应用[J]. 武警医学, 2023, 34(3): 185-189. XIE Xiaolian, DU Tao, QIAO Jing. Application of LASSO-logistic regression model in the analysis of influencing factors for hyperuricemia. Med. J. Chin. Peop. Armed Poli. Forc., 2023, 34(3): 185-189.

链接本文:

http://journal08.magtechjournal.com/Jwk_wjyx/CN/ 或 http://journal08.magtechjournal.com/Jwk_wjyx/CN/Y2023/V34/I3/185

[1]	王东军, 张颖, 田之魁,等. 国内高尿酸血症临床指南方法学质量与报告质量评价[J]. 中国中药杂志, 2022,47(2):547-556.
[2]	张啸, 赵薇, 胡焕青,等. 基于group lasso logistic回归的6月龄婴儿贫血预测模型的构建[J]. 中国慢性病预防与控制, 2020, 28(3):199-204.
[3]	武杰. Logistic回归中的随机Lasso方法[D]. 北京:北京建筑大学, 2018.
[4]	李敏捷. Lasso-Logistic与Group Lasso-Logistic模型在出生缺陷研究中的应用[D]. 太原:山西医科大学, 2016.
[5]	柴秀俊. 基于高斯混合聚类的切换系统的辨识[D].乌鲁木齐:新疆大学,2021.
[6]	李文政, 顾益军, 闫红丽. 基于网络贝叶斯信息准则算法的社区数量预测研究[J]. 数据分析与知识发现, 2020,47(9):90-96.
[7]	左俊, 李坤, 邓大同,等. 2017-2020年成人体检人群高尿酸血症患病率调查分析[J]. 安徽医专学报, 2021, 20(4):10-14.
[8]	Wang H, Shi L, Liu C, et al. Association between uric acid and metabolic syndrome in elderly women[J].Open Med,2018,13:172-177.
[9]	邓笑伟, 吴凡. 4004例军队干部高尿酸血症患病情况及相关因素分析[J]. 中华保健医学杂志, 2013,2:96-98.
[10]	Nurshad A,Rasheda P,Shahnaz R, et al. Prevalence of hyperuricemia and the relationship between serum uric acid and obesity: a study on Bangladeshi adults[J].PLoS One,2018,7(1):1-12.
[11]	李晓梅, 厉晓伟, 刘宇亭,等. 东南沿海雷达部队官兵高尿酸血症患病情况调查及其危险因素分析[J]. 空军医学杂志, 2021, 37(2):122-123,155.
[12]	张文然, 王猛, 汤朝阳,等. 男性γ-谷氨酰转肽酶和肌酐与高尿酸血症之间的关系[J]. 江苏医药, 2021,47(7):701-706.
[13]	张立军, 谭冰洁. 基于迹lasso模型的变量筛选方法[J]. 系统工程, 2018, 36(12):127-134.